]\gt ; [ 285 ] VIII .
JIathematicat Contributions to the Theory .
On Principle of Homotyposis and its Relation to Heredity , to the the Inchv idnal , and to of , Roce .
T. egetable kingdom .
By KARL ARSON , F.R.S. , With of ALICE LEE , D.Sc .
, RNEST WARREN , D.Sr .
, FRY , D. FAWCETT , B. Sc. , Reeeived Octol ) 6 , \mdash ; Read Novcmbcr 1 1900 .
CONTENTS .
I. TRODUCTORY .
( 1 .
) ener Scope of IsIemoir ; relation of undifferentiated like , . . . . .
286 II .
THEORY .
( 2 .
) Discussion of the Relation of Fraternal Correlation to the Correlation of like Organs .
, . . .
) III .
OBSERVATION .
}surement of Variability and Correlation of undifferentiated like in the Individtlal .
( 3 .
) Methods of selecting and collecting Material . . . . .
291 ( 4 .
) Statistical processes employed in reduction of Material . . .
292 ( 5 .
) Terminology adopted : and , . . . . .
293 IV .
ACTUAL DATA .
ection I. Leaves of Trees . . . . . . .
) ( 7 .
) A. The Ash\mdash ; Buckinghamshire , Dorsetshire , Ionmouthshire Scries . . . . .
294 ( 8 .
) B. The Chestnut\mdash ; Buckinghamshire and lixed Series . . . . . . . .
300 ( 9 .
) C. The Beech\mdash ; Buckinghamshire Series . . . . . . . . . . . . . .
( 10 .
) D. The Holly\mdash ; Somersetshirc 1tnd Dorsetshire Serics . . . . . . . . .
304 oction II .
Segmentation of Seed-vessels of Plants . . . . . . . . . . .
( 12 .
) A. The Wild Poppy hamshire { omersc B. The Shirley Poppy\mdash ; Chelsea and reitt Hampden Sories . . . . . . . . . .
) ( 13 .
) C. Hispanicn\mdash ; Slough Series . . . . . . . . . . . . . . . .
( 14 .
) Hampden Series . . . . . . . . . . . . . .
322 ction III.\mdash ; Members of Whorls .
oodrnff ( Series , ( 294 ) 12.11 .
] 901 .
K. ARSON AND OTHFR ON Page Section .\mdash ; ( 16 .
) Ferns . . . . .
326 ( 17 .
) A. rtstongue ( \mdash ; Somersetshire Series 327 ( 18 .
) B. Ceter Series . . .
331 Section Veins in mios .
( 19 .
) The Onion ( Allinm ) Great Hampden Series 333 Section VI .
rminous Plants .
Seeds in Pod .
( 20 .
) The Proom \mdash ; Yorkshire ries .
( 21 .
) Iushroom ( ( 22 .
) The Wild Ivy . . .
OF ( 23 .
) Relation of Homotypic and Fraternal Mean Correlatio]s . . . . .
Relation of Individual and Racial Variabilities . . . . . .
Possibility of Relation ) etween Homotyposis or Heredity and ility .
No evidence of changing intensity of any of these with tionary Descent .
360 ( 26 .
) Brief of Conclusions reached . . .
, . .
362 \mdash ; On the Homotyposis of the essels of Leguminous . . . .
364 I. INTRODUCTORY .
( 1 .
) THE present paper endeavours to deal with a problem upon which I have long occupied , adopting the widest basis compatible with the time and means at my disposal .
In the first place , I have often been impressed with the small reduction in variability which can be produced by selection .
The offspring of a single parent while ging in character , possibly very widely from the average character of the race , will still have a variability in that character only slightly reduced , say at most 10 per cent. below the racial variability .
Even if we select the ancestry for an indefinite number of generations , the offspring will have a variability upwards of 89 per cent. of that of the original race.$ Now this capacity in the parent for producing variable offspring must be in some manner related to the degree of resemblance in tbose offspring .
We have thus the two fundamental divisions of our subject : What is the ratio of individual to acial variability ?
How is the variability in the individual related to inheritance within the race I must endeavour to explain my meaning a little more fully and clearly .
The individual puts forth a number of like organs , corpuscles in the blood , petals of the flower , leaves of the trees , scales on the wing .
These may or may not be divided up into differentiated groups .
Special forms ofleaves occur in the neighbourhood of the fruit ; florets may be differentiated according to their position on the flower , scales ' ' On the of Ancestral Heredity 'Roy .
Soc. Proc vol. 62 , p. 400 , HOMOTYPOSIS IN THL VEGETABLE KINGDOM .
according to their position on the wing ; there may be two or more classes of blood( , orpuscles .
But if we take gaus which are like , and so we can judge undifferentiated , there is still a difference between them .
The individual in putting forth what I may perhaps term " " undifferentiated like organs , \ldquo ; does so with certain measure of variability .
I am not concerned at present with the soul.ce of variability .
It may be due to the individual enyironment , to the physique of the individual , or to some inherent bathmic tendency in the individual due to its ancestral history .
All I am concerned with at present is , that the erentiated like organs of an individual possess a certain variability , and that this bility is somewhat less than that of all like organs in the race .
If , however , the bility were entirely due to external circumstances attending growth , we should hardly expect to find leaves gathered from different anches and on different sides of more alike to each other than to leayes of other of the same race .
But this is indeed the case ; there is a considerable correlation among the undiiferentiated like organs of an individual notwithstanding the small reduction on racial val.iability .
Among the trees and plants considered in this paper we have data enough to enable us to determine whether , say , a hundred leaves placed before us were gathered a single tree , or collected at random from a numbel of trees .
Here , of course , I by to " " mine \ldquo ; to state the odds for or against these two assumptions .
Just as we can find by the methods already discussed in earlier memoirs of this series , the degree of correlation between brothel .
S and the val.iability of an of brothers due to the same parentage , so we can determine the correlation , , the degree of resemblance between the rentiated like organs of the individual and the degree of variability within the individual .
This determination is the answer to our first fundamental problem , that of the ratio of individual to racial variability .
But turning to the ) rocess of reproduction , the offspring depend upon the parental germs , and it would thus seem that the degl.ee of esemblance between offspring fully conscious of verbal difficulty in the " " tiated like ) I ) the distinction between the differentiated and the is quite either from the udpoint of or from that of the frequency distribution itself .
Differenti tion , hither d ) to function , position 011 the individual , of productio1l , , is connected wibh one or ell-mrked dominating factol .
S ; it is discoverable the ) utionfol heterogeneity .
On the other hnnd , bility in " " likc \ldquo ; is not to any one or two dominatiug factors which be it is dne to that comI ) of small causes , , which leads to what is .
iu both ) homogeneous distribution .
A diversity due to } erentiati diability due to are quite distinct .
The oue is the esult of dominting factors which can be isol described ; the other of a great number of small factors , varying from organ to organ pable of being defined or specified .
Indeed , upon each dominating factor of differentiation is snperposed such it chance ariability .
Of course all things which differ even by chance variation are in senso differentiated .
But the term tion is throughont this ) reserved for tho differoltccs , not multiplicity of msob , fium ) influences .
, 1901 .
] must ) on the val.iability of the sperm cells and the ova which may each be fairly considered as " " undifferentiated like gans.\ldquo ; Here again we are not compelled to assert that much or little is due to environment and little or much is due to inherent ancestral influence .
All we assume is that such causes as produce the likeness between leaves of the tree , or florets on the same flower , produce the likeness between spermatozoa or of the same individual , and that on this likeness the ultimate esemblance of from the same parent depends .
We have then to investigate how the quantitative resemblance between offspring of the same parents is related to the quantitative esemblance between the undifferentiated like organs in iudividual ; and then we must test on as wide a range of data as possible this theoretical elationship .
Now the reader will perceive at ollce that if we can back the resemblance of of the ents upon the esemblance between the undifferentiated like of the individual , we shall have largely simplified the whole problem of inheritance .
Inheritance will not be a peculiar feature of the reproductive cells .
One , let us say , differs another in that it produces blood corpuscles more or less alike and unlike those of another frog .
In the simplest forms of reproduction , buddin and parthenogenesis , the offspring will not be absolutely alike , for buds and entiated like organs , and such organs have only a limited degree of lesemblance .
If this view be correct , variability is not a peculiarity of sexual duction , it is something peculiar to the production of undifferentiated like organs in the individual , and the of heredity must largely turn on how the resemblance between such organs is modified , if modified at all , by the conditions of nurture , growth , and environment generally .
Our discussion of the subject will naturally break into the divisions : On the theoretical relationship between the correlation of offspring and the correlation of undifferentiated like organs in the individual .
A determination of the elation of entiated like wide of life as possible .
The present paper deals only with variation and elation in the vegetable dom , but the data variation elation in the animal kingdom being collected at the same A comparison of the degree to which the results obtained from ( b ) satisfies the obsel.ved values of correlation already found when we use the elatio determined in .
II .
ICAL DISCUSSION OF ILLATION OF lATERNAL COltRELATlON TO CORRELATION OF JFERENTIATED LIKE ORGANS .
Let be deviation the mean of the general population of any indiwith to any character .
Let us suppose to depend upon certain HOMOTYPOSIS IN VEGETABLE KINGDOM .
characters in the spermatozoon and certain characters in the ovum from which the individual has developed .
These characters cannot of course be determinerl , still less measured , but we have no reason to doubt their existence .
In the ticular spermatozoon from which the individual has deyeloped , let them have deviations . . .
from their mean values for all the spermatozoa of race , and let . . .
be the corresponding deviations for the ovum chal.acters .
Then , . . .
where is a quite unknown function .
The mean of the -character will , however , correspond to the mean values of the spermatozoon and ovum characters , and if we suppose the variation of these characters small as compared with their mean value , we assume as usual such deviations : . . . . . .
, where the 's and 's are independent of the 's and , and define the iJld female inheritance .
Now let be the standard deviation of the character in the population ; the standard deviation of of Let be the correlation of , and of and .
Then we will suppose that there is no selection of particular by particular spermatozoa , or that and are not correlated .
Then if number of individuals in the population : where is the sum for all indiyiduals of any or for constant subscript , and 2 is the sum of and for every possible subscript .
This follows by simple squal.ing and remembering that .
We thus reach : .
Now let us consider the correlation of two individuals due to the and ova put forth by the same two individuals .
Let and be the values of their characters , and represent the fundamental characters in the two ) matozoa , and two ova on which they depend .
Then we have .
Now let us multiply by and sum for every fraternal pair ; then if be the VOL. CXCVII.\mdash ; A. PLOFESSOR K. ARSON ANY ) OTHLRS ON we have , since one set of will have the same standard tion as the other , if they be taken of the same sex , where is the umber of pairs of brethren .
Hence with alne notation as before and the same assumption as to no selective process between the female and male reproductive cells .
Now are undifferentiated like organs put forth by an individual .
If be the correlation of such : Simil In the next we should have all of brothers : We thus deduce . . . . .
Now if the degree of resemblance between undifferentiated like organs vary largely from organ to organ , we could proceed no further with this investigation .
We should shown that there was a relation , namely ( iii .
) between the and the , but as the characters in the spernlatozoa and are unknown and unmeasurable , we could not proceed further .
On the other hand , if the co1Telation between undifferentiated like organs is approximately constant , then we may replace correlation coefficients like and by their mean value , say .
We have next to consider such correlations as or .
These are the two correlations between two different characters in two spermatozoa or ova from the same individual .
Such correlations must vanish if there be no organic correlation between characters of the same spermatozoon or ovum , ( ii .
) if there be no correlation between the reproductive cells put forth by the same individual .
In other words , must contain and as factors , and since if these two correlations are perfect must be perfect , it is a reasonable hypothesis to assume . . . . . .
This , it will be , is identical in form with the result I have sllpposed to hold IN KINGDObJ .
for cross-heredity , and which appears to be approximately true for that case .
We shall consider later statistics bearing on this result .
Assuming accordingly that holds , we find from ( iii .
) by ( ii .
) .
Hence we conclude that .
Or : Tlte correlation between brothers ) be equal to the correlation the rentiated like organs forth by Now , if this result be true , it is very remarkable and very fundamental .
We should hardly expect it to be absolutely true , for it is very unlikely that the coefficient of correlation between undifferentiated like organs is the same whatever the organs may be .
This equality may rather be spoken of as belonging to an ideal theol.etical vital state approaching the actual state , perhaps , as BOYLE'S Law or the perfect fluid approach phenomena obseryed in physical nature .
What we should expect would be a general approximation between the values of , and a tendency to equality when large series are averaged .
This is the point which shall investigate in the sixth section of this paper , after placing before the reader in the fourth section a ] wide range of actual statistics .
It will be seen at once that if epresents an approximate or average truth of living nature , then the problem of inheritance is to a large extent the same as the problem of variability in the individual .
We have not answered , of course , eve in part , the problem of why two brothers each other , but we fsee that it is part of a much more general problem having nothing to do with sexual reproduction .
It is one with the problem of the likeness in leaves of the same tree , or the likeness in scales on the same spot of a moth 's wing .
It brings the of heredity into closer touch with the problem of variability .
When we ascertain the sources of variation in the individual , then we shall have light on the problem of resemblance .
III .
ON THE VARIABILITY AND CORRELATION OF [ TNDIFFER ] ] LIKE IN THE INDIVIDUAL .
( 3 .
) I must frankly admit that I have collected my material from the standpoint of the mathematical statistician and not of the trained field aturalist .
have things which were easy to count and measure , and endeavoured to avoid " " differentia ' The of Heredity ' Roy .
Soc. Proc vol. 62 , p. 411 .
Not widc we regard the ge of living from the of the ) ' that been spent on the collection .
tion\ldquo ; only by the roughest tests and by largely untrained powers of observation .
I examined my material to see if it gave obvious signs of heterogeneity in the tabulated bulk .
I have examined individuals and their organs for obvious outward signs of differentiation .
But I have not studied the morphological eyolution of the considered , or questioned whether the parts counted were all due to the same source .
Hence it is possible that the botanist may reject at once some of ny series .
What I have endeavoured to do is to take as wide a range of as different organs as possible in different types of life and trust to the bulk of my statistics to give me a substantially accurate average value of to compare with the values of we have determined on other occasions .
At the same time the material here presented does not by any means exhibit all the trials made ; we often enough at the first attempt did not get a suitable character to measure or count , or again the individuals collected were occasionally insufficient .
Nothin , however , has been omitted was unfavourable to the conclusions ultimately drawn .
Thus the statistics have been retained , although there is not the slightest doubt of a very considerable differentiation in the flowers growing at different parts of the plant .
When such a differentiation takes place result will generally be a great reduction in the correlation ; for " " like\ldquo ; organs on the same individual , say A and if differentiated , will really be less closely related to each other than to and respectively , the corresponding organs in a second individual .
On the other hand a heterogeneity of material , say a mixture of two different local races , will tend as a rule to raise the correlation , for it generally amounts to compounding two very like corl.elatio surfaces with the mean of one approximately shifted along the regression line of the second .
If we bear these two opposing influences in mind , differentiation tending to reduce , heterogeneity to increase the actual degree of correlation of undifferentiated like organs , it will not seem incompatible with an actual aJpproximate constallcy of such correlation to find a fairly wide range of values in our statistics .
We set them forth as the first rough attempt to appreciate the resemblance of like parts within the individual .
To the specialist in the future must be left the work of selecting , with special knowledge , truly homogeneous material and absolutely undifferentiated characters , and thus obtaining the required correlation to alnuch higher degree of accuracy .
( 4 .
) It remains to explain the process by which the correlation was deduced .
Let us take as an illustration beech-leaves .
One hundred trees fairly of the same age and belonging to the same district , were selected , and twenty-six leaves specified by the letters of the alphabet were gathered from each of these .
The leaves were gathered so far as possible all round the outside of the tree , roughly about the same height from the ground , and scattered over different parts of the individual boughs .
Thus each tree was supposed to be individualised by twenty-six leaves .
The veins on these leaves were then counted , and varied for beech-leaves in general between ten and twenty-two .
Sce ' Phil. Trans , vol. 192 , p. 277 .
HOMOTYPOSIS IN ETABLL KINGI ) All the possible pairs were now taken , i. e. , in number , and entered on a correlation table in the usual manner , the two variables bein the number of in the first leaf and the number of veins in the second leaf .
But as either member of the pair might be a " " first\ldquo ; leaf , the table so formed was rendered symmetrical by starting with either leaf in the pair as first or second .
Thus a single tree led to 650 entries in the correlation table , or with 100 trees there were 65,000 entries .
This large number must not lead the reader to overweight the importance of the constants calculated upon it .
There were only 100 trees leading to 32,500 pairs of leayes , each pair coming from tree .
I should have much preferl.ed a thousand trees , but the great labour of collecting , counting , and calculating precluded any such number .
In many cases also it would have been practically impossible to have obtained 100 indiyiduals growing under fairly like environment .
For xalnl ) I only in approximatin to 100 Spanish chestnuts from one district and near one age .
When I took a second hundred , gathered from several districts , partly old trees and partly others of a pollard growth of fifteen to twenty years , I found the heterogeneity at once increased the correlation ( see below , p. 301 ) .
In many cases , of course , it was impossible to obtain twenty-six undifferentiated like organs from the same individual .
In these cases the pairs were formed in the manner , but in some series the total number due to each individual varied very considerably , and accordingly the work of verifying the tables was much increased .
When but few pairs could be obtained from each individual , we have.sometimes increased the number of individuals dealt with up to a coul ) of hundred .
But the of dealing eyen with a hundred indiyiduals is often\mdash ; for example , in the case of mushrooms and onions\mdash ; very serious .
Had it not been for the generous help of a number not only of willing but of competent collectors and calculators , the here dealt with would taken not eighteen months but years of my own unaided efforts .
The calculation of the means , standard deviations , probable errors , and correlations of each table was carried out in the manner sufficiently discussed in earlier papers of serie The symmetry of the tables leads to slight simplifications in calculating the product moment which will readily suggest themselves to the reader , and of course only a single mean and standard deviation is required for each table .
Tests for the accuracy of the last two constants are at once proyided in the case in which the same number of organs are taken from each individual , for thei values must be identical with those obtained for the whole series of organs entered only once and not for each possible pair .
( 5 .
) The quantitative measurement of the degl.ee of resembl nce between undifferentiated-like organs being , so far as I am aware , a quite novel ) of tion , I venture , with some hesitation , to introduce cel.tain terlns to oft* See especially Memoir III .
, ' Phil. Trans , pp. 263-318 , eluoir I Phil. Trans , vol. 191 , pp. 229-311 .
ideas .
I call undifferentiated like organs homotypes , indicating that they are types produced by the same mould or individual .
Thus two leaves of the same tree , two blood-corpuscles from the same frog are homotypes ; their resemblance will be homotypic , and the character by which their resemblance is quantita- tively measured will be the character .
I shall speak of otypic correlation and distinguish it from correlation\mdash ; although , of course , the former is in a cel.tain sense organic .
By organic correlation I refer to the correlation between two different in the same organ ; by homotypic correlation to the correlation between the same or different characters in a pair of homotypes .
lf the nIle character , then the correlation is dircct .
lation ; if different , then correlation is cor relation .
astly , the principle that homotypes correlated , i. e. , that variation within the individual is less than that of the race , or that undifferentiated like organs have a certain degree of shall speak of as homotyposi Thus homotyposis denotes not only likeness of the homotypes , but that this likeness has probably definite quantitative limits .
If my view be correct : heredity is only a special case of homotyposis , any multiplication of cells is homotypical , and denotes a given degree of variation and a of likeness .
This does not " " explain\ldquo ; heredity , but shows it merely as a of a much wider natural process . .
ACTUAL DATA .
SECTION I.\mdash ; Leaves of ( 6 .
) leaves of trees provide lllaterial for fairly easy computation without measurement .
Thus we may count the yeins on the leaf or the leaflets on the compound leaf , and ascertain the degree of resemblance betweell leaves of the same tree .
The variation is , however , in some cases very considerable , and the labour of forming the tables involving thousands of entries very great .
Still the leaf-series are some of my best , and considered as a whole , perhaps the most satisfactory .
( 7 .
) A. Ash ( Fraxinus excelsior ) .
Number of Pinnae Leaf.\mdash ; My first series consists of 26 leaves taken from each of 109 trees by Dr. ALICE LEE or myself The trees were nearly all large old trees , growing on the commons of Great and Little Hampden , hamshire .
The leaves were taken as far as possible a variety of branches all round the tree , and from different points on these branches .
The labour of tabling and deducing the constants is due to Dr. LEE .
The number of pairs My second series ( ii .
) consist of 26 leaves taken from each of 120 trees by Miss 1 ) .
FAWCETT , B.Sc. These trees , many of which were young trees growing in the hedgerows , were taken in the neighbourhood of Lyme Regis in Dorsetshire .
rules as to gathering were observed .
Dr. LEE undertook the of tabling and ths deduction of the constants .
The nulnbel .
of pairs 78,000 .
IOTYPOSIS IN THE VEGETABLE KINGDOhJ .
My third series consists of 26 leaves from each of 100 young trees being included , but the same rules as to being observed .
The whole of the labour of gathering , counting and deducin the constants is due to Miss MILDRED E. BARWELL , of Girton College .
* The general numerical results are given in the table below:\mdash ; Resemblance of Ash Leaves from same Thus a total of 329 trees , giving 8,554 compound leaves counted and yielding 213,850 pairs on the symmetrical tables , were dealt with in all .
It will be seen at once that the mean number of to the leaf , and their yariubility differs considerably with the locality , the age and the enviroment of the tree , but the correlation of leaves from the same tree is fairly alike for the series , and almost identical with the theoretical value given by the Law of Ancestral Heredity for the degree of resemblance between pairs of brothers .
The following is the actual distribution of the frequency of the leaflets:\mdash ; This was remarkably satisfactox.y piece of work , carried through from the collecting of the material down to the calculation of the constants , the statistical methods ving to be stcred applied during the course of the investigation .
In determining the probable errors of the correlation coefficie1lts , there is some to whether we should use the number of individuals , which seems too small , the number of which seems too or the number of individual organs dealt with .
I have used the latter , but enclose the errors in brackets to mark the doubt .
It will be remarked at once that the leaves with an even number of leaflets are relatively infrequent ; there is usually a single leaflet at the end of the compound leaf .
I was able at Hampden , however , to obtain leaves illustrating almost the whole evolution of this single leaflet .
Sometimes this leaflet had the slightest division at its point ; this division was found in most stages of double tongued leaflet down to a complete double leaf , which would be reckoned as two pinnoe .
In the same way centre leaflet was found slightly trisected , three tongued and finally a triple leaflet scarcely distinguishable from the usual single leaf at the end and the two ] ' side leaflets .
Very occasionally the double leaflet seemed to give one side leaflet and the final leaflet .
Of course , all these anomalies were rare and had to be sought for , but they would form in themselves a suggestive study .
The Dorsetshire series gives 9 instead of 11 as the mode , and 7 are more frequent than 13 .
This latter result is also true for the Monmouthshire series , which again has nothing like the same preponderance on its modal 11 as the Buckinghamshire series .
I take it that the greater variability of both Dorsetshire and Monmouthshire series , together with their greater relative proportion of compound leaves with 9 and 7 pinnoe , is due to those series containing a much larger proportion than the Buckinghamshire series of small trees .
For ashes in general , I think , we may safely take 10 as the average ( but not the much vl ) number of pimlaj , the standard deviation is pinnoe , and the leaf 1esemblance is measured by a correlation of 4 .
If we compare racial and individual variabilities we have : Thus 92 per cent. of the total racial variability of this character of the ash is to be found in the individual tree , and here , as in all our other series , it is impossible to assert that variation is a result of sexual reproduction ; it is clearly an essential factor of the individual growth .
I conclude with actual tables of data for the three series .
( 8 .
) B. Chestnut ( Castanea vulgaris ) .
Veins in the Leaf.\mdash ; A character which first sight appears easy to determine is the number of vein running from the main rib to the of leaf The leaves of many trees , however , will on examination be found unsuitable , the veining bein far too complex , too much indeed of a twork .
Among irly practicable leaves are those of the beech and the Spanish estnut .
Here two fundamental difficulties occur : near the tip of the leaf the number of veins becomes rather difficult to register .
In the case of the beech by bolding the leaf in various lights or even using a lens , it was possible to be fairly sure of the separate .
In the the chestnut , the total number of veins is so great that the addition or omission of a single vein is hardly likely on the average to lead to serious .
( ii .
) The occasional branching or bifurcation of the vein formed a difficulty , especially if it occurred near the edge of the leaf If the bifurcation occurred close up to the main rib of the leaf , the vein system was counted as two ; if close to the edge of the leaf as one ; diate bifurcations had to be left to the judgment of the recorder , having regard to the general appearance of the leaf and its system of subsidiary veining .
The doubtful cases were not very many , and occurred more frequently with chestnut than with the beech leayes .
In neither case was the main rib or included the total number for the leaf , the record was of the side veins flowing from this .
The numbers right and left in the beech leaf were found to be far more nearly equal than in the case of the chestnut , which thus possessed considerably greater asymmetry .
In these cases as in those of the ash , the leaves were gathered at random round the tree , and at random on the branches , care being , however , taken in the case of the chestnut to avoid leaves in the neighbourhood of the fruit .
Series .
Chestnuts .
I had hoped to get records for 26leayes from each of 100 trees from the splendid chestnut avenue at mpden House , but it provided records for some 60 trees only .
These I owe to Miss LINA cKENSTEIN .
I was able to add by diligent search about 25 additional trees from the same neighbourhood .
All these were large forest trees of splendid growth .
In order to make the number up to 100 , some 15 additional trees were added , these being all large old trees from other neighbourhoods .
I look upon this series as a fairly homogeneous one , representing the degree of resemblance between leaves on fully developed forest trees .
The counting of the veins on this series was undertaken by Miss C. D. FAWCETT , B.Sc. , and the ulating and numerical reduction by Dr. LEE .
Both these processes involved an immense amount oflabour , in fact days and days of careful and fatiguing work .
leaves were gathered in the autumn , tow the fall .
HOMOTyPOSIS IN THE KINGDOM .
Series ( ii .
) .
Mixed Spanish stnuts .
My second series consists of 26 leaves from each of 104 trees .
These were gathered in a variety of neighbourhoods .
Many of these were young In the case of nearly 50 sets , the trees were pollards from old stumps with about 15 years ' growth ; some of the others were from fine old trees , and as a result we have many districts and many ages combined .
I have to thank Miss C. D. FAWCETT for sets from the neighbourhood of HamI ) stead , Miss for sets from that of St. Albans , Miss CHARLES for sets from the neighbourhood of while about half the series was gathered by Mr. ROBERT J. PARKER from his chestnut woods at Haslemere .
The counting of the vein was Miss M. NOTCUTT , while the tabulation and numerical reduction are the work of Dr. W. R. MACDONELL .
The series being mixed , the variability is than that of the first series and thus the labour spent on it even The following table giyes the general results:\mdash ; We note at once that the mean number of veins for the forest trees and for the mixed group is sensibly the same .
On the other hand , the variability and the correlation differ very considerably .
The latter is sensibly higher than the 4 obtained for the leaflets of the ash .
I should be inclined to consider that the heterogeneity of the mixed series is at the bottom of difference between the two series , but there cannot , I think , be any doubt of the greater resemblance of tnut leaves for this character than of ash leaves in the number of pinnaj .
The actual distribution of veins in the 5304 leaves examined was as follows : \mdash ; The total gives a fairly smooth series with the mode at 40 , and a secondary mode at 42 .
In the forest trees we have the same result as in the total , but in the mixed group the secondary mode is at 44 , and almost equally important with the primary .
The influence of symmetry makes itself felt in the preponderance of the even numbers , although not in such a marked degree as in the case of the ash , where the counting of the final leaflet gave , of course , preponderance to the odd numbers .
For the Spanish chestnut we may accordingly take 41 veins as the average , 40 as the modal number to the leaf ; as the S.D. of the yeins ; and 5 ( slightly weightin the homogeneous series ) as the degree of resemblance between the leayes of the same It remains to consider the comparative individual and racial variabilities : see that the variability of the chestnut leaf as measured by the coefficient of variation is somewhat less than that of the ash .
The individual as compared with the racial variability is also somewhat less than in the case of the ash , but the individual exhibits still some 85 per cent. of the racial val.iability .
Tables IV .
and V. give the data for the chestnut series .
HOMOTYPOSIS IN VEGETABLE KINGDOM .
( 9 .
) C. ( Fagus sylvatica ) .
Number of Veins on have here only one series of 26leaves from each of 100 trees .
These were and counted by Mrs. KARL PEARSON and myself in the neighbourhood of Great Hampden .
The trees were in part fine isolated specimens growing on the commons , but as there were not sufficient of these we were compelled to gather from the somewhat smaller trees abounding in the beechwoods of this part of Buckinghamshire .
The leaves were gathered about 4 to 5 feet from the ground on the outside of the tree .
The tabulation of the results and the calculation of the constants are again the work of Dr. LEE .
Accordingly , we see that the modal value of the beech leaf veins is 16 , which is very near the mean , .
The standard deviation about , and the of resemblance among its leaves Measured by its of variation , it is the least variable of the trees so far dealt with , and the leaves of the same tree more closely resemble each other .
Notwithstanding its relatively small bility , the individual on an average exhibits about 82 per cent. of the racial yariability .
The table of reduced data is given below .
* I did not observe among the leaves gathered anything of the differentiation indicated by Profcsso ] hIACLEOD between ' licht- en schadernbladen The materi , so far as this cter ( number of veins ) goes , seemed singularly homo eneous and of small variability .
( 10 .
) D. Holly ( Ilex Aquifolium ) .
Number of Prickles on the .
\mdash ; This character seemed an easy one so far as mere counting goes , although the leaves were troublesome to gather .
The leaves were gathered , so far as possible , all round the tree or bush , about the ight at which cattle would browse , and from the outside of the bush .
Internal and high branches ( often differentiated by being almost prickleless ) were to be ayoided .
Flowerin trees , and , aboye all , leayes near the flowering buds or fruit were to be rejected .
These conditions could only be approximately followed in practice when a large number of trees had to be dealt with from one neighbourhood , and these series , I fancy , allow more to the personal equation of the gatherer than is wholly satisfactory .
Series .
SomersetshirHollies .
I owe this series of leaves from each of 100 trees to Miss AGNES FRY .
Of these trees , 56 were hedge-row hollies , and 44 wood trees or bushes growing in moist places .
* The counting of the prickles was in some part undertaken by Mr. MACLEOD , * Miss writes to me : " " The texture of the leaves seems rather different , \mdash ; the wood hollies are much less stout and harsh and more oflen have few prickles .
In fact , I rather wonder if does not affect the number .
In the few cases where hedge-row hollies have very prickles the leaves come from the inside of bush .
I have avoided the neighbourhood of berries , \mdash ; and what is more difficult of flower buds .
But the holly has two flowers , one with abortive pistil and one with aborbive stamens .
The latter I suppose would drop away and leave no trace of their presence .
But , on the whole , I have avoided flower and berry-bearing bushes\mdash ; it is easy to find bushes without either .
And , so far as my ervation goes , I am not convinced that the neighbourhood of flowers or the top of the tree causes modification .
It is very difficult to keep to a standard of counting for the tip , but I have tried to do so .
Doubtless the hollies too are of rather different ages , but I have avoided patriarchal trees , though doubtfud if this makes much difference in this case It will be seen that in this point , as in others , much had to be left to the judgment of a cal'eful observer .
HOMOTYPOSIS IN THE KINGI ) but more than three-quarters of the trees by Miss FRY herself The labour of tabulation and the deduction of the constants was by Dr. LEE .
Series ( ii .
) .
Ifollics .
This series of 26 leaves from each of 100 trees was hered c counted by Miss C. D. , B.Sc. It differs in several important respects from Miss FRY ' series .
AB it from the bourhood of the , by Lyme Pegis , we ]night expect a rable dence of ; the trees were chiefly hedgerow trees .
They show , howeyer , a ) mean number of prickles than the series , which embraced a large element of wood trees .
I think the is homogeneous than the omersetshire s. In contributes the almost abnormal series of with 1 and 2leaves with 2 prickles .
The lump at the end of the frequency ) ution is due to one .
I do not feel able to reject it entirely , but I the ical constants as detellnined with and hout it .
I take it that of seryers the ) ersonal equation of one tended rather to emphasise need for a tically " " random selection\ldquo ; of trees : and the personal of other to emphasise the importance of ically " " homogeneous group\ldquo ; of trees .
The result has been increaQe of correlation and variability in the first , their decrease in the secolld case .
Probably their is not far the truth .
It is noteworthy that the ash trees from Dorsetshire and Monmouthshire , although exhibiting a like substantial difference of environment , were far more The tabulation and eduction of series is also due to Dr. LEE .
I now summarise the results as before:\mdash ; In round numbers acr , the number of prickles ] the holly leaf , the standard deviation prickles , and the degree of .
The holly thus falls between our results for the ash and one series mshire ashes , the the gronl ) of nixed somewhat in excess of the amshire eeches .
V0L .
CXCV I K. AND ON The actual quency of pi.ickles is given The is thus seen to be fairly uniform , the mode corresponding in the total result as well as the Somersetshire series to 15 prickles , while that of the Dorsetshire serie 3 is at 14 , an even number of prickles .
The holly leaf has clearly not the marked symmetry of ash or chestnut leaves .
Thus , 16 pl.ickles in the Somersetshire sel.ies are almost as frequent as 15 , or since there is almost always a prickle at the tip , asyt.ical leaves must be very nearly as frequent as symmetrical .
stly , to the comparison of and individual variabilities , we have the table : \mdash ; Ifere again we see ] great is the variability the individual .
Looking as so largely biasing complete Dorsetshire series that we must exclude it , we see that the possesses 80 to 90 per cent. of the vftriability of the race .
The yariability of the offspring of the individual will , of course , be larger still , even if we had merely uniparental reproduction .
Thus we have still further evidence , if more were , that variability is not a factor dependent upon sexual reproduction .
For this character of the prickles it will be seen that the holly leaf is more variable than those of the ash , chestnut , or beech .
The following are the fu ndamental tables of data:\mdash ; IN . . . .
( 11 .
) ving dealt several in iety ( it to deal with in iety of tces in lcll of the yegetable kingdom .
with my colleague , ssor F. O. , the nati c psules of llpies were selected , they h au easily cter , which could be triety of races .
there son diversity iu the nlethod of ection .
The ules 5iven being that the bands to be counted of the capsule and llot at the ] ; thus ( ii .
) evely possible bud the plant to be .
The went on spring to autullm of the ; it was impossible to collect , for counting all the wild poppies soon they had ) .
some poppies were when they had done otlJers with flowers on them , and others with lnany buds .
Few , I think , would be included which not formed their buds .
my own series , I counted the on every capsule , in flower , and in nearly every bud , ) the bud and using a ) microscope .
The laboul of this was , and was not out by all helpers .
I doubt , deed , whether any advantage was derived il , in so far as there may be the numbel of bands the or late deyelopment of the seed-vessel .
( 12 .
) The following obtained:\mdash ; Series .
Wild Rhoeas .\mdash ; I ecimens from as wide dside , rerow , thd , and counting every set ) down to the bud I could nlanage .
These specimells the very top of the Chiltel.ns ; the soil is not very my mean numbel of buds small .
I myself bled the calculated the constants .
Series ( ii .
) .
.\mdash ; Miss C. D. gathered 236 poppy plants on the southern slope of the Chilterlls , They nearly ffi from single lfield , and thns their much more ulliform , and the nate mable t in Series .
It is quite possible that these largely the ) of comparatively ) as they gathered the calculated the constants .
( iii .
) .
) .
\mdash ; Tho of were gathered by Miss the tntocks , but SG , I added to it several other in SOlt K. ON AN1 OTHERS ON counted various less active hclpel .
S. Thus the series is really sn omnibus inly f the ntocks .
tabulated the data calculated the My two series ) of the , a and beautiful garden variety .
Series .
.\mdash ; These phnts were grown by my colleague , Professor F. O. OLIVER , in his gardell at Chelsea .
The plants had scarcely been thinned , so , very crowded , grew tall , and with few buds on the plant .
The ants were taken up and to me by rail , but being delayed in nsit suffel.ed a deal , so that a good many psules were broken off , or could not be counted .
These circumstances account for the smallness of the number of pairs obtained from each allt .
OLIVER kindiy came and assisted me in the counting .
The data were tabulated reduced by myself Series .
Poppy.\mdash ; These plants grown at Hampden House , and covered a large border .
They were extremely healthy , large I have counted as many as 120 capsules on a single plant !
The were taken up singly as they had done flowering and the bands counted .
labour of counting was sevel.e , as there were 4443 capsules on 176 plants as compared with 1020 capsules on 325 Chelsea poppies .
The tabulation involved the of 197,478 pairs , pel.haps the largest number we have yet dealt with in a single correlation table ; this and the determination of the constants is the work of .
LEE .
The reader will notice that the vestigations on poppies differ largely from those on the leaves of .
In the latter case , 26 leaves were takell at random from the and the chal'actel .
on each measured .
In the former case , eyery ayailable pStlle was dealt with .
These , of course , are not all the capsules put forth by the plant , and the of these two sets would vary considerably from one series to another , approaching close to equality in my wild poppies from the top of the Cbilterns , and differing in the ) from Chelsea .
In the leaves of the trees tree weighted by its size or vigour ; in the case of a orous 1 , howevel , it may times as many pairs to table as a feeble plant .
is a good deal to be said for both methods of dealing with the problem , and even a method to which I shall refer immediately .
If We pick up two leaves andom and wish to consider the that they belong to one tree , we ought certainly to allow for the larger numbel of pairs of leaves on the ; if we fish to discovel to ) what extent Titndomly chosen men have chal.acters like to those of others , we ought certainly to allow for the ilies , even if we do deal with possible in those families .
hese two different metbods of with resemblance due either to the ] ) oduction of like by the individual or to heredity must be carefully borne in mind .
In dealing with Mr. GALTON'S data for stature*I have taken every possible pair of 'Phil .
Trans , vol. 187 , pp. 253-318 .
HOMOTYPOSIS IN TABLE KDOhI .
brothers ; in my own measurements of 1000 families I have taken only two brothers each family .
We cannot expect the results to come out exactly the same .
Indeed , we might , priori expect the latter method to give rather higher results than the former , for it introduces individual families , and if all the brothers of one family only or all the like organs of one organism , we should find nothing but spurious correlation growing indefinitely small as the number of brothers or organs was indefinitely increased .
Thus I anticipated that the Hampden Shirleys would give a lower relation than Chelsea Shirleys , the wild poppies from the top of the Chilterns a higher value than those from the bottom , \amp ; c. As our series of poppy plants were in no very large , it was impossible to directly test this result by taking only a single pair from each plant , eyen if it had been possible after the , and marking to select a pair of capsules at random .
I adopted a third or diate method .
liss FAWCETT had corded her capsules under the letters , \amp ; c. , and not grouped them directly under the number of the stigmatic bands .
, it was possible to take with with , \amp ; c. , and thus repeat no single capsules , but a nnlnbel of pairs sensibly taken at random .
Thus large and } ) lants give llore pairs than feeble ones , and yet much fewer than if we take every possible This series will be referred to as Series .
Its general result confirms anticipation referred to aboye .
It will be noticed nothing of this kind can be done in the case of leaves of trees , where we cannot gather ) leaf One great advantage of the 26 leaves from every tree is the means it provides of accuracy of the process of tabulation and calculation .
Each leaf is used 25 times , and thus the mean and standard-deviation of the total number of pairs to be the same as those of the total series of leaves ( say , 2600 for 100 trees ) .
This valuable method of verification is lost when we do not take the same nunlber of like organs from each individual .
It then becomes of interest to tain how far the mean and val.iability of the total nulnber of organs are identical with the mean and variability of the total number of pairs of organs observed .
This comparison is made in Series , the mean and variability of the 4443 capsules being compared with those of the esponding 197,478 .
TIre whole of the tabulation and arith1netic involved in Series is mine , but is again due to Dr. LEE .
now proceed to sum up results obtained from the above seven series .
worked out the theory of this some time , and it occasion OFESSOR K. PEARSON AN ] ) OTHERS ON Resemblance of Poppy Capsules from the same Plant .
Now this table contains a good deal of interest .
Upwards of a thousand poppy plants were dealt with , involving the counting of the bands on more than ten thousand sules , and the entry of more thaJl a qual.ter of a million of pairs in the correlation tables .
We see at once that the mean and variability of the Shirley poppy differ so widely from those of the wild poppy , that it is idle to group these species for these chal.acters together .
It is interesting to note , howeyer , that both in the number of bands and in the yariability the Hampden poppies , whether wild or Shirley , stand at the top of their respective series .
With regard to the wild poppy , I probably gathered from a wider area with a greater variety of environment than was the case in Series ( ii .
) or ( iii .
) .
That mine were gathered from uch poorer soil is evidenced by my plants having on an average only 6 capsules , while the Quantocks and southern Chiltern poppies had about 10 capsules .
The difference is possibly eyen greater than these numbers indicate , for I counted eyery bud I possibly could .
The two Chiltern series do not differ much in the mean number of bands , but the difference ofyariability is very ificant .
I should take it to be a very fair measure of the relative range of environment in the two .
When we turn to the Shirley series the erence is almost too great for any coml ) arison .
While the Hampden series had some twenty-five ) ) sules , on an average , to the plant , the Chelsea ries numbered * The readcr must ) that this ; the of of the clolV , or the nunll ) of HOMOTyPOSIS IN THE GETABLE KINGDOM .
only about three !
In fact , besides being extremely crowded a process of selection had gone on in the transit .
From examination of the uninjured plants , I think that five capsules a plant would be more than the ' of these poppies before xcking , and that at least one to two capsules per plant were destroyed on the journey .
Accordingly , I do not consider that much weight hold be given to the results of this series , but I felt bound to include it as I have included all the series dealt with , even when I hold them to be really unsuited to the present investigation .
If we consider the two methods of dealing with Series ( ii .
) , i.e. , that of ling all possible , and that of including pairs in which no capsule was repeated , we find that no very sensible made in the mean or variation , but a rather greater increase in the correlation than might have been anticipated .
I think this is largely due to unconscious selection in the pairs , , we took the records for and ( b ) , ( c ) and \amp ; c. , as our pairs ; it is quite possible that in recording the psules successive letters would correspond to capsules gathered from the same part of the plant , and so the members of the pair would be subject to the same or somewhat similar conditions of nutriment , light and warmth during development .
Valuable investigations have yet to be made on the relative degree of resemblance of leaves or seed-yessels coming at random from the same tree or plant , and coming from the same branch or shoot .
The effects of position on the individual undoubtedly produces some differelltiation , and all we can hope to do at present is to get round numbers approximately describing the degree of reselnblance of these undifferentiated like organs .
If we exclude the Chelsea series , we have for the Wild poppy 4983 Shirley poppy 5238 Mean . . . .
5110 For the four series of trees we have Ash .
Chestnut . . .
Holly Beech . . . . .
3918 Mean . . .
4984 So far as the seed-vessels of poppies and the leaves of trees are concerned we have a degree of resemblance in undifferentiated like organs which approximates to the value CXCV1 .
PROFESSOR K. PEARSON AND OTHERS ON It will be as well to put together the freq have done in the case of the trees:\mdash ; These distributions appear to be uniform although asymmetrical , and they give no obvious sign of a mixture .
The mode is 10 for the Chilterns series and 8 for the Quantocks series , while the Shirley poppies give 12 for Chelsea and 13 for Hampden .
It remains to tabulate the several results obtained for variation .
These are given in the following table .
We see at once that the general effect of weighting the capsules with the number of pairs on the plant has the effect of increasing the mean and reducing the variability ; in other words , the plants with most capsules on and leading to most pairs , are those with the greatest number of stigmatic bands .
We have some evidence accordingly that the most vigorous exhibit most hands on the capsules:\mdash ; ND OTHFR ON So that the variability of both may be expressed in round numbers as 16 .
It is clearly not a constant for the whole vegetable kingdom , but 16 enables us to form some appreciation of the plasticity of vegetable forms , and may be compared with valueF obtained from other series later .
Comparing the ratio of the racial and individual biJities , we see that the latter reaches , on average , over 85 per .
of the former , the mimmum value and the maximum 91 .
It is difficult to test this point directly , for the very simple reason that a single plant will , as a rule , have very few flowers ; and upon small numbers it is impossible to test satisfactorily the full variability of the plant .
The probable error becomes enormous if we have only four or fiye capsules to determine the standard deviation from .
The method adopted has been that which arises in the treatment of parental inheritance ; all the offspring of parents of a given stature are clubbed together and form array .
This array has a certain variability , and this variability is taken to be that of a single family of offspring having parents of the given .
It would clearly be impossible to find the variability inside the family numbers such as occur in a single human family .
To some extent we can verify the assumptions thus made .
Thus I took the wild poppy with the largest nuInber of capsules I had come across , and determined their variability ; there were 3 such capsules , and the standard deyiation was .
I then took two of the most Shirley had 126 and 85 capsules on them , and their yariabilities were given by standard deyiations of and a mean of .
These are actually in excess of the standard deyiations of the arrays ' .
namely , of the two series of poppies !
I thin we may accordingly take the latter as a reasonable measure of the possible variation of the individual .
PROFESSOR K. PEARSON AND OTHERS ON ( 13 .
) C. Hisprxnica.\mdash ; Besides the poppy capsules , Professor OLIVEP .
had suggested to me that it would be possible to count the segmentation of the seed vessel in the cases of Nigella and Mallow .
I accordingly procured from Messrs. VEITCH , of Slough , upwards of a hundred plants of gellcx H. Meanwhile , ofessor O sent me a list of some he had counted himself , and pointed out that there was a substantial differentiation between the seed vessels on the main stem and those on the side shoots .
Tt was quite clear on examination that these seed vessels were not undifferentiated like organs , the number of segments in a seed vessel on the side shoots bein much more like the number on a similar vessel on a separate plant than like those of a main-shoot capsule on the same plant .
Further , although I do not know the history of these special plants , they were probably from highly selected seed and in a very unstable condition .
Whereas in the examination of many thousand poppy capsules I only found one remarkable abnormal case , abnormalities in the case of Nigella abounded and rendered the counting of the segmentation very difficult .
Double and treble capsules were found in state of union up to complete fusion , and with every variety of deformity , such as the tips of one capsule growing through the sides of a second capsule .
Thus the material was by no means suited to my present purpose .
However , as I had the material it seemed worth while testing the influence of differentiation in reducing correlation .
Accordingly I made no attempt to separate top from side capsules , and formed the following table in the manner sufficiently indicated in the preceding pages:\mdash ; * A double capsule on one stem .
It contained only a few seeds .
These Professor F. O. OLIVER kindly owed for me this year ; here resulted two plants with a very few poor flowers , but the mality not perpetuated , and the capsules rotted off without giving seed .
If the top capsules only } ) taken I should often have bad only one o11 a plant .
PROFESSOR K. PEARSON AND OTHERS ON While the modal value is conspicuously 8 , the mean value is .
This is due to the second modal value of 5 on the side shoots .
The standard deviation is and the correlation .
It will be seen at once how the result fully confirmed my expectation , that the degree of resemblance between capsules on the same plant would be largely reduced , owing to the existence of differentiation .
The actual distribution of capsules in 210 plants was as follows:\mdash ; Thus on an average there were four seed vessels to a plant .
The mean and standard deviation found directly from the capsules are and ) respectively .
These are in substantial with the above results found for pairs .
They do not confirm the conclusion based upon the data for poppies , namely that if we work with pairs we shall raise the mean and lower the variability .
The fact is that vigour seems to work in opposite ctions with this plant and the poppy ; with the latter it means many capsules with a high number of stigmatic bands , but with the former it means lnany side shoots with the differentiated capsules with segmentations .
Such capsules of course decrease the mean and raise the variability .
The coefficient of variation is , sensibly greater than that of the wild poppy , but below that of the holly leaf The standard deviation of an array is , or the variability of the individual 9 per cent. of that of the race .
The plant with the greatest number of seed vessels on had only 11 , and these gay a standard deviation of , greater than that of the whole race !
But of course this is subject to a very large probable error ; it suffices to ennphasise , however , how the source of variation is to be sought within the individual , and is not a mere result of sexual reproduction .
The following table sums up the results obtained ( 14 .
) D. ( Malva Rotundifolia).\mdash ; I had given up the idea of the ordinary wild mallow ( Malva sylvestris ) owing to the few plants to be found in the neighbour .
Hampden , as well as on account of the immense number of seed vessels on the ants to counted .
But one day an inspection of my potato-patch revealed IN the whole ground covered with low phnts of in found that a cottage had recelltly stood on the site , and that the cottagel .
S kept in their herb-bed a small clump of this plant for the preparation of a local ointment of high repute .
When the cottage was removed the mallow had spread over the whole patch , I believe very little by seed , but largely by stolons .
This seemed to provide excellent material ready to hand , and from it we can draw some conclusions as to the influence of such a division on the degree of resemblance of like parts .
It will be remembered that DARWIN , in on vulgar , found that cross-fertilisation of flowers on plants which had separate roots , but had spread by stolons from the same individual , presented no advantage over self-fertilisation , and he concluded that this arose from the fact that the plants were really one and the same individual so far as such were concerned .
If this were so in the case of my mallow , we ought to find a great reduction in the relationship of undiflerntiated like ol.gans .
Two points struck me at the outset\mdash ; the small variability in the segmentation of the seed vessels and the low mean value compared with that ( 15 ) given in botanical works .
I was able to collect 127 plants ; of these , 76 had more than 26 capsules , some many more , and of these I took 26 only ; 51 plants had fewer than 26 capsules , and although I took buds in every stage of development I could not make up my number to this , often only obtaining 10 , or even fewel .
As the potatoes were being removed the mallows had to be taken up the seeds were ripe , and the seed-vessels artificially dried .
The counting of the segments proved a very laborious task .
To Miss M. NOTCUTT I owe a considerable portion of the counting .
The buds and small seed vessels requiring a lens or small microscope were counted by Dr. LEE , Mr. RADFORD SHARPE , and myself I am responsible the tabulation of the data and for all the numerical calculations .
The 127 plants gay rise to 57,740 pairs distributed thus : .\mdash ; Malca of in the OFESSOR K. PEARSON AND OTHERS ON The following table gives the chief numerical results:\mdash ; These results amply bear out my conclusions .
The degree of resemblance of like parts has been very largely reduced .
The variability is smaller than any we have yet come across , and the variability of the individual differs under 2 per cent. from that of the race .
All these conclusions are compatible with the decreased individuality which may be expected to exist among the members of a group to a large extent propagated by stolons and not from seed .
Although in the majority of cases ( 76 out of 127 ) I took the same number of seed capsules ( 26 ) from the plant , still the influence of the plants with fewer capsules is quite marked ; we see that weighting with the number of pairs lowers the variation and raises the mean , in other words , the large vigorous plants have more segmentation and less variability .
The distribution of segmental frequency is fairly uniform , and shows a marked mode at 14 segments , not the 15 of the usual statement .
SECTION III.\mdash ; Members of ( 15 .
) Woodruff ( Asperula odorata).\mdash ; I must confess that there is divergence between the standpoints of the botanist and of the statistician .
To the latter a is good or bad according as it affords facilities for fairly easy measurement or enumeration .
He has first to seek such characters as a sine non , and then inquire how far they occur in undifferentiated like the other hand , the will readily provide a list of such characters in plant or animal life , and not HOMOTYPOSIS IN THE LTABLE KINGDOM .
$25 one of them be available for measurement or counting .
I had long fixed my eye on the members of the whorls of the woodruff They were so easy to collect and count ; they exhibited such a moderate amount of variation , and one felt sure the collecting , counting , tabulating , and calculating could all be done in the inside of a week ; and the impulse to such a light task after the elaborate work on the earlier series was very great .
Professor F. O. OLIVER and others wal.ned me that the members of the whorls were differentiated in their origin*and also in their position on the stem , and that such a series was unsuited to illustrate the degree of between undifferentiated like organs .
However , the task seemed , comparatively speaking , so easy that I felt I would undertake it , if only to compare with the Nigellct Hisp , and ascertain in another case how diffel.entiation does weaken the degree of resemblance of like parts .
I accordingly collected 201 single sprays of woodruff well distributed along about a mile of lane on the bank of a hedgerow facing south .
These were gathered at Great Hampden , Buckinghamshire .
In counting the members on the whorls I soon found eyidences of differentiation in position , the whorls towards the top of the spray having , as a rule , fewer members than those lower down .
, The following is the table of frequency of different pairs of whorls :\mdash ; *Some of the members of the whorls are trne leaves and others modified stipules .
They broken off as close to the as possible ; they all parts branch .
off above grou1ld , but the forked sprays were only few in mber . .
PEARSON AND ON Woodrufl .
The double modes 6 and 8 are brought out by the aboye distribution , and this is evidence either of rentiation or of a tendency to an even number of members .
Contrary to the usual rule , the mean number of members is reduced when we take pairs , indicating that when we take a spray with many whorls\mdash ; such as occasionally arises forking\mdash ; the nullber of members in the whorls are reduced , and since the variation is also very slightly increased , they are spread out over a range .
The great in the correlation , the value of which is only , shows how unsuited the material was for the purposes of the present investigation .
At the same the atio of the to the racial variability reaches the very high value of !
We have selected a character to test individuality on , which exhibits ces which gely racial , and not peculiar to the individual plant .
SECTION ( 16 .
) In choosing ferns for considering the resemblance between undifferentiated like organs I had in view not only the comparative ease of counting ( as compared , for example , with the segmentation of mallow ) , but the fact which students of the fern strongly essed upon me that it was peculiarly subject to its environment .
One great authority went so far as to assert that the presence or absence of sori on an individual hartstongue depended on the environment and not at all on the individuality of the plant .
While hardly prepared to accept to the full such a statement ( for I had already learned in a variety of types of the existence of individuality in a lllarked manner apart from the influence of environment ) , I HOIsIOTYPOSIS IN THE VEGETABLE KINGDOM .
anticipated in accordance with it fairly high correlation of undifferentiated like organs , nd this has , indeed , turned out to be the case .
( 17 .
) Somersetshire Hartstongue ( Scolopendrium yulgare).\mdash ; I owe the counting of the sori on 8 to 12 fronds of each of xtstongue ferns to Miss AGNES .
It was not always possible to obtain the full suggested number of 10 fronds .
In a few cases Miss FRY also counted more than 10 .
But the ferns with fewer than 10 are not many , and the distribution of the fronds is sensibly identical with that of the pairs of fronds .
The number of sori was often very large , running up to 180 ; the grouping of the leaves was accordingly arranged in units of 10 sori , and , to save fractionising in the table , these groups were taken , 95\mdash ; 195 , 195\mdash ; 295 , \amp ; c. The grouping on the basis of Miss record was carried out by Mr. LESLIE -MOORE .
With regard to the counting of the sori , all the sterile as well as the fertile markings were included .
Fronds with no markings were included ; one plant , the only one found , with absolutely no markings on any of the fronds was excluded , and to this I shall recur immediately .
It was difficult to procure ten perfect fronds ; many of those gathered were found to be more or less torn ; in many cases the loss of sori was easily ascertained ; in others an estimate had to be formed .
Such estimates were made , however , in only forty-four cases out of more than 1,000 fronds , and in the majority of these cases it meant the addition of very few sori , and a possible error in the estimate of still fewer .
Sometimes one or two sori may have been missed at the tip , where the small markings are indistinct , but the errors due to this source as well as to the occasional error of a unit or two in counting such .
a large are well under the unit of grouping of ten sori finally selected .
The environment of the series was not quite uniform , some thirty-three coming from a stream bed the remainder from a park glade .
The plants varied considerably in age , but the necessity of procuring eight to twelve fronds was probably effective in excluding any quite young ferns .
The absence or presence of on the frond was not directly associated with its age or size .
Miss FRY has recorded a number of large and medium fronds with no , and some of the fronds recorded as young have occasionally a very considerable number of sori ( as many as ninety-five on one at least ) .
Still , a glance at the table on p. ,330 shows that the fronds with zero sori form in same way an lous group ; they give a hump in the frequency distribution in a manner similar to that due to barren wives in fertility-frequency curves .
Dr. LEE having tabulated the sori , and calculated the constants on the basis of including all fronds with zero number of , I went through Miss FRY'S data again and excluded , not all the entries in the first row and column of Miss LEE'S table , but only such of them as arose from the fronds with no sori .
The result was a great smoothing of the table , a raising of the mean , and a lowering of the variability ; but no effect whatever was on the degree of resemblance of like organs , although I * See Chances of Death and other Studies in Evolntion , ' vol. 1 , essay on " " Reproductive Selection and HOMOTYPOSIS IN THF , TABLF I We see from the last ] how the " " hump\ldquo ; disappears from the lency d we exclude the fronds without .
The double node i least considerable flatness at the top of distribution , ] ) some heterogeneity of material .
I am vever , to ) variability ( about forty-two even when the fronds without ) is fully accounted for by the sensitiveness of the fern in the matter of.$orl to very slight differences of environment .
I look upon the high resemblance of like parts here as having been intensified by this cause ; an inherent lity whave superan individuality due partially , perhaps , to age , but to small of immediate environment .
Although the variation in the race is so , and the degree of individuality so great , the ratio of individual to racial bility is still 78 per cent. Thus while the yariability of the hartstongue ( as far as sori is concerned ) is double that of the very variable holly , and almost that of oppy , still the percentage variability of the two latter species is to that of the former only about 85 to 78 .
Thus in the most variable and most individual species we ] yet come across , we still find the variation within the individual is more than of the entire variation of the race .
In view of facts like this , it seems impossible to maintain the position taken up by Mr. ADAM , that variation is the outcome of bi-sexual reproduction .
The source of variation exists within the individual is extensively active without the occurrence of any form of mating whatever .
I place here the table the distribution of ) in pairs of fronds .
in brackets are those which must replace the )bers , if fronds sori be excluded .
VOL. CX ( !
VIf.-:080 K. AASON ANY ) ON OTYPOSIS IN THE ) ( 18 .
) B. Somersetshirc , ( Ceterach officinarum ) .
time after I had tried in to coumt any characters in bracken , I had letter Miss AGNES saying that she thought it just possible that the lobes on the fi.onds of ceterach might be counted .
The chief difticulty , of course , the indefinite character of the lobes near the tip\mdash ; a difficulty which had rendered the lobes in bracken impossible .
The rather indefinite tip is here of greater consequence ] lall in the hartstongue or the chestnut leaf , for the total number of lobes is comparatively small .
In some samples Miss FRY sent me , however , I agreed closely with her estimates , and although the tip must form a difficulty , ' we settled that an attempt should be made to include ceterach in the present series .
Miss FRY accordingly counted 9 to each of 99 plants .
It was necessary to take plants of very different ages , and eve if ceterach be not so sensible to its environment as hartstongue , it is quite possible that part of the elation observed is due to similarity of age .
The absolute agreement of the result obtained with that for hartstongue is one of the most striking things in the whole collection of data .
Considering the difficulty of the tip in ceterach and the .
sensibility of the sori of hartstongue to environment , it may be a chance agreement , but it is certainly one that gives ground for pause , and suggests further investigations of the degree resemblance between like organs in ferns .
The tabulation of the data and the calculation of the constants are again due to Dr. LEE .
Ceterach .
The distribution of frequency of the fronds is almost the same as that of the pairs , ce the attempt was made to take 10 fronds ; actually 80 plants Lad 10 , 14 had 11 , and 5 only 9 fronds .
Frequency Pairs .
* Miss writes : ' It is very difficult to keep to of nting for the tip , ) I trierl to do so OTYPO , IN VLGETABLE It will be seen at once that while in the case of rach the bility has fallen to less than half that of the close to that of the Nijclla and less than that of holly\mdash ; still the measure of the idual i of sameness in like organs , is to the of accuracy defined by its probable error identical with that of hartstongue .
Considering the difficulties isiug f environmenb , difference of age , and from standards of eckoning , this reement is remarkable .
It seems to me that the basis of the identity , the bulk of the agreement , must have other sources than a balance between the resemblances produced in undiflerentiated like organs by such diverse causes as ellvironment on the one hand and on the other acting upon the individual planfs .
I hope that further investigations upon ferns may be forthcoming .
Table XIX .
contains the classified data for ceterach .
V. ( Allium cepa ) , ) ( 19 .
) The veins in the successive tunics of onions occurred to me as a possible character for computation , no differentiation of the number of veins with the position of the tunic being discoverable .
The number of tunics on which the yeins were countable was about six , but the process of counting itself was for more than one reason a most trying one .
As there were so " " like \ldquo ; to individual , it was needful to deal with at least 200 .
I accordingly purchased about 250 English onions ; these were in three groups , and came from as many cottage gardens .
Their ayerage horizontal diameter was about 2 inches , and they appeared very uniform in character .
It is quite possible , however , that there may be some eneity introduced into the series by this mixture of onions sources , but I did not at the time see my way to the whole series from one homogeneous sowing .
* The entire work of counting and on 200 of these onions most yaliantly undertaken by Dr. LEE .
Each onion cut through ally ; the tunics then came easily apart , and these were held to the light , and the number of veins , ranging from 18 to 54 , then counted and ecorded .
The following table contains the chief numerical constants:\mdash ; * The Ilions were ] ) of formed a large chancel crucilix of spccios co Tho nglish onions from this crucifix a most eIlGOtlS g ) uniformity of size and .
I these by the of dler ' ' offerings ) thom .
If any of my readers wish to the of the , they the cffect of slicing and counti11g or five ouion , and bhey will , I ) ined , I soon , to give the task despair .
K. ON ' Onion .
The actual number of tunics with veins countable was made up as follows : 1 onions with 6 , 44 with 5 , 34 with 4 , and 1 with only 3 available tunics .
It will be that the onion has a degl.ee of resemblance in its undifferentiated-like organs closely resembling that of the Chelsea poppies , of the Dorsetshire hollies ( without No. 91 ) , or the mixed Spanish chestnuts , and somewhat less than that found for the two series of ferns .
Its variability proaches closely that of the wild poppy .
On the whole it fits well into our series of results , the somewhat high value of the correlation } ) ) those of several sets , the homogeneity of which is not entirely above ) .
The accomp nying table gives the classified data from which the constants were deduced .
gumino uPlants .
( 20 .
) I had at an early stage considered that the counting of seeds in the pods of leguminous plants would be a very easy manner of testing the intensity of homotypic correlation .
But as number of seeds fully deyeloped would depend on the extent to which the flower had been fertilised , it seemed to me , after further thinking about the blem , that unless I counted all the abortive as well as the fully developed seeds , I should find the factor of homotyposis weakened by the external chances upon which I thought rtilisation must depend , and which I feared would have nothing in the main to do with the individuality of the plant .
I made some attempts to count all abortive as well as fully deyeloped eeds in certain species , but I the task not only very laborious , but my estimates doubtful .
Examining for another purpose DARWIN'S ' Cross-and Self-Fertilisation of Plants .
' I was struck by the fact that he did not hesitate for the ) urpose of comparing their vigour to count the seeds of plants living in the open and olilce subject to free ( 7 isitation from insects and bee He distinctly states that " " the difference in the number of the contained seed must depend up the constitution of the plants Thi.s view of the matter impressed me without entirely removing my earlier doubts .
a number of plants under the like collditions , and with a superabundant visitation from insects , there would be ample and equal chance of fertilisation for all DAISWI in cases to the good or fully developed seed .
HOMOTyPOSIS IN THE VEGETABLE KINGDOM .
of them , and excess or defect from the average number of seeds would depend only on the constitution of the individual plant .
It would seem accordingly that in the case of cross-fertilisation ample visitation was a , and this led me to select broom in the first place .
Broom ( Cytisus Scoparius ) , Dale.\mdash ; We collected 10 pods from each of 120 broom bushes .
These were growing within some quarter of a mile of each other , on the roadside at Botton , towards the head of Danby Dale , Yorkshire .
I roughly estimated that if abortive seeds were to be included , 16 would be about the modal number of ovules ; the actual average of fully developed seeds in 1200 pods was , with a modal value at 9 , so that some 6 to 7 seeds failed of fertilisation on the average in each pod .
The following is the equency distribution of the seeds in the pods : I came across no pods with an entire absence , although such ) exist .
I refer to this because the reader might think that such were discarded .
The tion is fairly regular , but it is clear that the variation is very large , the distribution being very flat topped .
As in nearly all the cases that I have dealt with in this memoir , it is markedly skew .
I reserve , howeyer , for the present the full consideration of variation in the vegetable world , as my data extend .
beyond the material considered in this paper , which is limited to the cases in which the homotypic correlation has also been worked out\mdash ; a far more laborious enquiry .
I would only remark that among plants and trees I know of nothing approximating eve to the " " normal law and that in many cases we appear to have Jllixtul.es of local races hardly yet differentiated by the botanist .
The following table gives the constants for broom:\mdash ; .
Seeds Pods .
The actual distribution of pairs is given the table on the } ) It will be seen that some of the alTays are rather irregular , but in its results I look upon broom to be as satisfactory as any material I have dealt with .
In the place , the table fully beitrs ( conclusion the simple HOMOTYPOSIS IN GETABL KINGDOM .
distribution of the high degree of variability in the broom pod .
We have so far reached nothing but the sori on the Hartstongue fern with anything like such a coefficient of variation .
We might expect , if Mr. ADAM SEDGEWICK'S view were correct and great variation meant small intensity of heredity , that homotyposis , of which heredity is only a special case , would also be small if variation were large .
But we find nothing of the kin .
The resemblance of undifferentiated like organs reaches the value fully equal to that of the pinnoe of the ash , which have than half the variability of broom pods .
This value is close also to the given by the Ancestral Law of Heredity for the degree of resemblance between brothers .
In the next place this normal behaviour as to the degree of resemblance of like organs is associated with a variability in the individual which amounts to upwards of 90 per cent. of the racial variability .
It is impossible to form standard deviations for groups of 10 , but if the reader will reduce any array of the above table to the number 10 , rejecting all fractions of unity , he will obtain quite fair samples of what I actually found in the case of individual pods .
I think feel confident that the variability of a race is not in inverse proportion to either its heredity or its homotyposis .
It will be seen that broom , so far as it goes , supports the view held by DARWIN that the number of ripe seeds is a measure of individual constitution .
At the same time the question of self-fertilisation arises .
Would completely self-fertilised plants exhibit full homotyposis ?
May not outside influences\mdash ; wind , shaking visits of insects or other causes of a random character\mdash ; be also needful for the of the seeds even in the case of self-fertilisation ?
Again , can we always suppose that a plentiful visitation will take the case of all cross-fertilised plants ?
I must confess that the value for the homotyposis found in the case of broom did not suffice to remove all my doubts as to any character depending on fertilisation being a suitable one for the determination of the intensity of homotyposis .
The subject , however , is one of such great interest that it deserves independent and fuller treatment than be provided here .
V. CROSS-H0M0TYP1C CORRELATION .
( 21 .
) ( Agaricus campestris).\mdash ; A more complete study was now made of two species\mdash ; the mushroom and the ivy .
We found some difficulty in discovering two easily measurable or countable characters in one organ , neither of which was largely influenced by the growth of the organ or the age of the anism .
In selecting an organ in the mushroom , I was guided by the desire to take a sinll ) and an organ upon which fertilisation had no influence .
The gill of the lJroom seemed to satisfy to some extent these conditions .
Had we been able to grow our own mushrooms we might have succeeded in taking them all at the same stage of This has been to some extent provided in Appendix to this memoir , added since its completion .
CXCVII .
3.38 K. ARSON ON development , and our results would in this case have been more homogeneous and thus more satisfactory .
But simply gathering them in the fields we could only make a very rough approximation to uniformity in age or development .
We aimed at a condition in which the underskin had disappeared and the gills had lost all signs of a crease or notch .
Dr. and I collected more than 100 mushrooms in the Great and Little Hampdens .
With the assistance of my wife we cut 10 gills out of each of these , distributed , as far as the condition of the mushroom would allow , uniformly round the axis .
These gills were then placed on ruled paper , the horizontal lines of which served to mark the maximum length of the gill .
This was marked by two fine needle-pricks , one at either end .
Two further needle-pricks were now made at the points farthest removed from this length lin , one on the upper and one on the lower contour of the gill .
Thus by drawing through the four needle points lines parallel and perpendicular to the paper-ruling we obtained a rectangle approximately circumscribing the gill .
length and breadth of this rectangle are what I shall term the length and breadth of the gill .
At first I endeavoured to take these measurements by the use of a micrometer microscope , but this method was very laborious , and gave an apparent exactness wholly out of keeping with the somewhat rough and ready manipulation of the gill .
Even when the gill itself was placed under the microscope the gill length could only be placed parallel to the spider wires by a somewhat vague appreciation , and the upper boundary of the gill was under the lens rather less definite than without it .
Accordwe resorted to the needle-pricks and the ruling by fine lines of the abovementioned rectangle .
I determined , after various trials with engraved glass scales , \amp ; c. , the length and breadth of these by the use of a pair of proportional compasses , set to multiply by ten , and an ordinary millimetre scale .
I was thus able to get readings to 01 of a millimetre , which agreed well on repetition .
As the millimetre itself was afterwards adopted as the unit of grouping , I do not think the errors of manipulation will seriously affect our results .
The needle-prick method enabled us to deal with the mushrooms quite fresh and reserve the measuring till another season .
Dr. LEE found what we may term the gill-index , i.e. , the ratio of length to breadth for the 1070 gills , by aid of a FULLER'S slide rule .
We had thus three characters to deal with\mdash ; length , breadth , and index .
Of these we considered that the first two would undoubtedly be largely influenced by the stage of development of the mushroom , but we hoped that the index ( like , for example , the cephalic index of children over two years of ) would not be much influenced by the growth factor .
On the other hand , the correlation oflengths of gills or the correlation of breadths of gills from the same mushroom would of course be influenced by growth as well as * In all such systems of mass-measurement , the method adopted has to be chosen with dne regard to time and eyesight .
The measurement of the gills pied me about ten full days , the microscopic method would have taken about four times the time , evcn if my eyes would allow of such continuous working and the mushrooms could have been preserved fresh .
HOMOTyPOSIS IN THE yEGETABLE KINGDOM .
33 !
) by individuality .
Thus we should find the degree of resemblance between two gills of the same mushroom , when measured by ngth 0 breadth , to be influenced not only by the indiyiduality of the mushroom , but by the of its growth , and accol.dingly likely to exceed , and possibly considerably exceed , the value about 5 , which we have determined in several other cases .
* In order to make a more complete study of the influence of stages of growth on the degree of resemblance of undifferentiated like organs , Dr. LEE with her characteristic energy undertook the five needfUl tables .
They are the following Organic correlation of the length and breadth of the same gill .
( ii .
) Homotypic correlation of the lengths of two gills of the same mushroom .
( iii .
) Homotypic correlation of the breadths of two gills of the same mushroom .
Cross homotypic correlation of the length of one , and the breadth of a gill of the same mushroom .
Homotypic correlation of the dices of two gills of the same mushroom .
Of these tables the cross homotypic correlation between length and breadth is a very laborious business , offering no means of readily testing the accuracy of the entries until the whole table is completed and its columns and rows added up .
Ultimately the frequency distributions ( as given by row or column marked totals ) for engths and breadths in tables ( ii .
) , ( iii .
) , and , ought to be ?
times the frequency distributions of lengths and breadths as given by the like row and olulnn in the organic correlation table , being the llumbel of organs measured in each individual ; in our case this is ten .
I begin by considering the results obtained for the degl.ee of resembla1lce of gillindices , i. e. , Table to above .
The following were the results obtained:\mdash ; Mushroom .
Gill Indices .
Pairs .
The frequency distribution of the gills was the following : \mdash ; * The absurdity of rding the ctor m ) realised if we consider how much the of resemblance in brethren would increased if we took of brothers , differing , \ldquo ; by less than two years , and these were all from foul ' to twonty .
PROFESSOR K. AND ON Gill Index .
This distribution is rather irregular may heterogeneity ; the tail is somewhat in extended .
But the numerical constants given in the table above are quite comparable with the best of our other series , and are rather better I had anticipated priori , for I expected the diverse stages of growth would influence the index as well as the absolute measures .
It would be of great interest to repeat the series on an artificially grown bed of mushrooms , taking each at the same stage .
It will be seen that the mushroom , with of resemblance of like organs about that of the Spanish chestnut , has a vari bility comparable with that of the ash .
The variation of the individual is even in such a comparatively simple organism more than 80 .
of that of the race .
It will thus be clear that the mushroom takes , so far as our present inyestigation is concerned , no place apart in the vegetable world , but quantitatively is comparable with highly developed organisms like trees .
The accompanying , Table XXII .
, gives the data for the measurement of the gills .
I now turn to the relationships of the absolute measurements , which , as I have noted , we have good grounds for believing are much affected by growth .
The numerical constants are summed up in the following table , the dimensions eing given in millimetres:\mdash ; HOMOTYPOSIS IN THE ) I. ,341 Absolute Dimensions of Mushrooln ] The results here given seem to be of erable suggestiveness .
While means and standard deviations differ very considel.ably , coefficients of variation for the length and breadth of the gill approximately equal , in addition very higl ] .
This is precisely what we might expect when dealing with of organisms in various stages of growth .
Ful .
there , the bility of the individual as ared with that of the race is reduced immensely below the 80 to ) cent. with which our vestigations have made us familiar .
This might crain be nticipated as a result of heterogeneity in the stage of growth .
The organic tion of length and breadth , as well as the homotypic tions of the same characters , both direct and cross , are all high , and the latter are much ) what we might legitimately put to the credit of the homotyposis factor .
It seelltS , therefore , somewhat hazardous to consider how .
they satisfy the relationship gested in the introductory part of this paper .
The direct ) lations f length and breadth are not approximately equal as we might expect ; this suggests that growth may be a somewhat more marked factor in the length the of the gill .
If we take the product of homotypic COl.relation with correlation , this ought , our hypothesis , to give us roughly ) ) correlation .
We have the following results:\mdash ; Product of direct homotypic length correlation with organic lengthbreadth correlation Product of direct homotypic eadth c organic breadth correlation Product of mean of direct motypic breadth and length correlation vitll organic length-breadth correlation . . . . . .
These numbers have to be compared with the cross homotypic lengtb-breadth correlation , i. e. , with6275 .
It will ) seen at once that they are sensibly too small ( say , 56 as compared with .
But having ecognised the influence of heterogeneity in the growth stages of our material , the divergence is , I take it , of a magnitude rather to confirm than coIlfnte the hypothesis as to the relationship of direct cross homotypic correlations .
It must be that we dealing with results .
Our main proposition is that ercdity is not a factor of life peculiar to sexual reproduction , but merely a phase of the factol , which we have termed homotyposis , or the tendency of the individual to put forth undifferentiated like organs with a certain of ance .
Our proposition is : that if the direct homotypic correlatio oscillates about a certain mean yalue , and the cross homotypic correlation the product of anic and direct homotypic correlations , then we should expect to find the average degree of resemblance of brethren equal to the average degree of resemblance of tiated like organs the individual .
All then I we can safely say for the gills of mushrooms is that the relation of the direct and homotypic lations is no means such that it condemns hypothesis , or enforces ns to reject our subsidiary proposition .
The difficulty lies in ding ( dodt onisms with undifferentiated like organs with two characters in sufficient quantity , easily couTlted or measured , upon which the hypothesis can be tested .
The investigation of the length and breadth of ivy leaves , to which we shall soon turn , suffers to some extent the same defects as that for the gills of mushrooms .
ables XXIII.-XXVI .
contain the length-breadth data for the mushrooms .
The uency distributions for length and breadth are given in the last row and column of Table XXV .
above , and exhibit in their irregularity something of the heteroeneity of growth to which I have referred .
Of the two tributions , I consider that for the breadth as the irregular and consequently the less satisfactory .
It is , of course , harder to determine a definite breadth for the gill than a definite length , and I should be well content to compare the product of the homotypic length correlation and the organic length-breadth correlation , i.e. , 6025 , with the cross homotypic length-breadth corl.elation , i. e. , 6275 , as the best basis for our subsidiary proposition available from these measurements on mushrooms .
( 22 .
) TVild ( Hedera Helix ) This series was originally undertaken by Dr. WARREN , his )ject bein to measure the degree of resemblance between ivy leaves on the same plant , using as his character the index , or ratio of length to breadth .
It was hoped that in this manner , the growth factor be irly well eliminated .
The ures ( page 348 ) indicate how the breadth and length were determined by Dr. in leaves of somewhat different shape .
It will be seen that , as in the case of the gill of the mushroom , the breadth is taken as the maximum breadth between tangents to the contour parallel to the length , and these tangents do not necessarily go through points like .
Twenty-five leaves taken from each of 4 plants by Dr. , in the hood of 3-l 8 K. ANY ) OTHERS ON and leaves f each of 58 plants by me , in the neighbourhood of Great Hampden in Buckinghamshire .
Care was taken to avoid the reproductiye shoots with their larger uncut leaves .
The Hampden nts were , I feel certain , a fairly homogeneous group , the plants were young runners up the trees , only a few feet in length .
The Canterbury plants appear to have had a somewhat larger leaf , although in themselves an apparently homogeneous group .
It would undoubtedly have been better to have had the leaves collected by one person from one district ; but until the leaves had all been measured and their indices found , I did not know of this difference between the two groups .
It was very little noticeable in the table of index ibution , but became manifest to me when finding the homotypic correlation table of the absolute lengths , which I did in two sections , one for either group .
The very considerable labour involved in measurin and determining the indices of 2500 leaves precluded a repetition of the work on more homogeneous material .
The whole of this labour was undertaken by Dr. WARREN , who further formed the table and determined the correlation coefficient .
* I give in the following scheme the frequency distribution of the indices to show its uniformity .
*After measnri the gills of only 1070 mushrooms , I very fully appreciate Dr. task on the IN Of course the groupings here are large , but the the mixture of seems more ular t in some of our other ries .
The following scheme gives the constants in the usual manner : Leaf Index of Wild Ivy .
It will be observed that the correlation is lower than we might have expected , although the mean of the index correlations for gills and ivy leaves , i. e. , , is almost identical with the value given brothers by the law of ancestral heredity .
The low value of the correlation gives of course a high value to the variation , the variation of the individual plant bein within 4 per cent. of the racial variation .
The riability of the ivy leaf , as judged by this character , has not the intensity which is popularly associated with it ; it is sensibly less than that of the mushroom gill , and about equal to that of the tunics of the onion .
I now pass to the absolute asurements of the leaf .
Dr. WARREN being much pressed with other work , kindly all his measurements at my disposal , and I proceeded to draw up the same four tables as in the case of the mushroom gills .
The work here was , however , much laborious as I had to deal with 25 leaves instead of 10 gills , and this involved 300 pairs for each plant , instead of only 45 !
In the course of a 's work I had completed the length-breadth organic elation , the length-length homotypic correlation , and about a third of the length-breadth homotypic correlation .
At this point .
LEE took the work off my hands and finished the last table and the breadth-breadth homotypic COl.relation .
So that the results are again the product of co-operation .
* I below the data arranged as in the case of the mushroom gills .
The dimensions given in eighths of inches .
We now systematised the vorking of these bless , .
10,000 to 60,000 providing appropriate checks for up to each of clion .
It to describe these here , but we be glad to put our expericnce the service of oue orkiug at similar problems .
At the same time the collection , measurement and formatiou of for the homotypic correIation of two characters in 25 undifferentiated like organs of 100 individuals fvill cost a worker at least three weeks 's fairly continnon bour .
K. AN1 ON of Wild Ivy Leaves .
These results very considel.ably from those for the gills .
The coefficients of variation even higher than those for the gills , and I think this is a result of a certain amount of rogeneity , as well as of the leaves on individual unners not being all quite the same stage of development .
In the case of the deciduous the leaves were gathered in the fall of the yeal , and no further development of vein was possible ; but in the ivy leaves , taken indeed at the same time , the equal deyelopment of all the leaves taken from the runmer could hardly have been ched when they were gathered .
On the other hand the correlations much lower than in the case of the gills , \mdash ; much more within the range of the results obtained othel characters .
I am inclined to think , therefore , that there has been some balancing of here , heterogeneity due to locus of collection and to stages of being to some extent counteracted by a differentiation due to position of the leaf on the runne ' I have not included these results for lengths and breadths of ivy leaves in my homotypic series because the values , although well within the ange of the other detel.minations , appeal to me to be somewhat fictitiously so .
The disturbing factors referred to above seem to me to have also given the cross homotypic correlation a higher value than it ought to have .
It will be seen that the correlation of length and breadth of pairs of leaves from the same runner is almost as high as that of the breadths of pairs of leaves from the same ru1mer .
I should expect a considerably greater inequality .
I regret that the great labour of these cross-correlation inyestigations has hindered their being carried further than these two cases in the vegetable kingdom , but in the zoological data , which we ) present in hand , I hope get material less open to Cl.iticism than in the cases of mushroom gills and ivy leaves .
These two series undertaken with the view of how far the use of indices would cancel the fluence of the factor of gl.owth .
As we have The must of coursc ) .
in mind the were taken from run1lers which had not reached the tops of their wnlls t1ees , , they not from reprodnctive shoots .
HOMOTYPOSIS IN THI TABLE K seen the mean of the index-correlations for mnshroom gills ivy leaves esult by no means bad , or widely the mean result of all our data .
If we test the hypothesis as to the value of the cross-correlation we ) the following results arranged as in the case of ooms : Product of direct homotypic length correlation with organic lengthbreadth correlation Product of direct homotypic breadth correlation and tnic lengthbreadth correlation Product of mean of direct homotypic length breadth correlations with the length-breadth correlations These numbers have to be compared with the cross lengt h ) readth correlation , with We see that , as in the case of the mushroom , they are somewbat too small , 48 say as compared with 52 .
But the difference is considel.ably less here , and allowing for the action of disturbing factors , I think we may say that the two quantities under investigation are at least of the same order of magnitude .
There I thin we must leave the hypothesis until my zoological measurements are reduced .
I give the four tables of classified data for the absolute lengths and eadths of the ivy leaves .
In the first or organic correlation table the reader will be able , in the distribution at least of dths , to find of the irregularity to which I have already referred .
VI .
SUMMARY or ( 23 .
) In summing up my results and them with those ined for fraternal correlation my co-workers and lnyself , I felt some difficulty .
If I made a selection of what I considered the best homotypic correlation series , and the best fraternal correlation , I might well lay myself open to the char$e of selecting statistics with a view to the demonstration of a theoretical law laid beforehand .
Accordingly , I determined to include all my homotypic results , except those for absolute dimensions of mushroom gills and ivy leaves , where it was pretty evident that we had to a greater or less degree an influence exerted by the owth factor .
I thus drew up Table XXXII .
, containing a summary of all my results .
I am quite sure that heterogeneity due to one or another cause , exerts influence of one kind at the top , and differentiation an influence of another kind at the bottom of this table .
The amount of these influences and of other disturbing causes , one cannot measure and allow for .
I can only hope that having taken a fairly wide range of races and characters , the influences to obscure the homotypic correlation , on the one hand by raising it , and on the other by reducing it , will about , and .358 ESSOR I PF , ATtSON AND OTHERS ON and the value for evity in man much below the true fraternal correlation .
In former case , mothers were few number , in the latter the non-selective death-rate reduces very considerably the intensity of collateral inheritance .
Both rreyity telnper ( included in this table for the same reason as woodruff in the first ) .
I would not run the risk of any apparent selection to reduce either homotypic or fraternal correlation to a closer range of values .
The meaJl of this second table gives the value for fraternal correlation .
Now I do not propose to lay yreat stress on what at first sight might look like a most conclusive equality between the valnes of homotypic and fraternal correlations , \mdash ; limits of the probable error ' ; 4479 and 4570 are indeed equal .
I am quite aware a few further series dded to either the homotypic or fraternal results modify to some extent this equality .
But what I would ask the reader to do is to examine the two tables side side , to note how the first and last several results of both Illay fairly be held to be subject to quite definite modifying factors , then to consider whether there is not very substantial evidence gathered from a fairly wide range of characters in nearly as wide a ange of species to show that both homotypic and fraternal correlation fluctuate about a mean value between and 5 .
will not venture to assert that either are absolutely constant , but I do realise that it is extremely difficult with the complex system of factors influencing living forms to reduce our conditions to theoretically perfect state in which we shall measure solely the factor we are .
If the intensity of homotyposis were exactly , I should be inclined to distrust any long series of results , one and all of which exactly .
There are so many other disturbing factors which only tho , who have endeavoured to collect series of this kind will fully appreciate .
In the first place , the theoretical conception of undifferentiated like gans is very hard to realise practically ; position of the organ on the branch or of the branch on a plant , howeyer careful be the collector , may really have introduced differentiation , and so weakened the homotyposis .
Secondly , the environmental factor comes into play .
It is difficult to obtain a hundred individuals with like environment ; soil , position with regard to other growths , sunlight , insect life , \amp ; c. , may differ in a manner that the collectol cannot appreciate .
Unlike environment may produce a fictitious likeness in the organs of the same individual when we pass from one individual to a second .
The fact that half our series grew in one field , the other half in another , that came from one side of a road , part from another , may introduce an unperceived heterogeneity which increases the apparent homotyposis .
Thirdly , the difFiculty of ensuring that all individua ] are of the same age or in the same stage of development , is very great .
The leaves of an old tree may have a fictitious likeness when compared with those of a young tree ; we may gather organs from one individual when it is in a stage of pment , which would only be reached in another individual some days or weeks later .
These and other factors may perhaps be to some extent eliminated\mdash ; far better of course by the trained botanist than by HOMOTYPOSIS IN THE ) the mere mathematician\mdash ; but I very much doubt possibility of their complete elimination .
If homotyposis had a constant value throughout nature , I should only expect this value to be ascertained as a result of ) eavel.age of many series in which the } ) ctors of rentiation , environment , ge , stage of growth , \amp ; c. , may more or less counteract each other .
In this nlanner we may approach a fair appreciation of the bathmic influence of individuality in the duction of undifferentiated like organs .
What I should accordingly deduce as legitimate the above general results would be this , the intensity of pure homo. .
typosis throughout the vegetable kingdom probably lies between .
; this is also the mean value found up to the for fraternal correlation .
We may accol.dingly conclude that heredity is really only a phase of the wider factor of honnotyposis .
At bottom it is only part of the principle that when an indiyidual puts forth undifferentiated like organs these are not exactly the same , but with a definite intensity of variation have a definite likeness .
When we associate heredity with sexual eproduction , we are only considering the of notyposis ( variation and likeness ) between individual ) between individual oya .
Such homotyposis leads to a likeness of the esulting from the zygotes , which has the same meall value as homotypic tion itself Thus if the continuity of germ cells between and offspring be ised , we face no longel the of heredity , but that of homotyposis , and this might possibly be rednced to the problem of or cell multiplication .
Why does the lifeunit on self-multiplication oduce h ) with a definite degree definite degree of variation ?
I hope for further light on this ) roblemwheIl my data for homotyposis in the animal kingdom , already being collected , are solnewhat more complete ; but only the biologist , not the 1nathematician , can solve it .
( 24 .
) Now let us turn to another point : holllotyl)osis involves , as we have seen , not only a certain degree of likeness in the group of homotypes but a cel.tai degree of variation .
Our series does not include any groups of than twenty-six homotypes , except in the case of the Hampden ) oppies .
Hence it is not lly possible to calculate directly the iability of the individual .
But from theoretical considerations , as well as from the support of individual instances , we heell that the standard deviation of the array is easonable m of the of the Of course direct investig tions 011 point ] be of great interest .
But I con cider that the present series indicate that 011 an the variation in the individual is some 87 to 88 per cent. of that of the race .
lf the reader will examine the column headed Percentage XXXII .
, he notice excluding the and breadths of in which the of growth is all important\mdash ; no percentage ) * It is quite easy to find } grcater of ility t( ovcn greftter variability thau that of the rnce , it is the ) iudividual is ) prosouted by the S. .
of the rray .
this seems to instructive when we consider the statements made by some who theorise about variation rather than actually measure it : The undifferenlike organs put forth by the individual in umbers 8to 90 cent. of the variation such in the race .
Does not this completely refute the views that variation is the result of sexual reljroduction , and that it is insignificant in the case of ?
Sexual reproduction may produce a type which is not that of either but this does not a fortiori alter the variability of the race with regard to any .
With continuous variation such type would have ) reviously existed as far as any special organ or character is concerned , its contributes nothing to the racial val.iability .
We have to meet the fact that the ) roduces undifferentiated organs with a remarkable degl.ee of variety , if the inyestigations of the present memoir be yalid the practical result of the homotyposis factor having a value of 4 to 5 would denote that the individual variability is to per cent. of the racial .
( 25 .
) we may considel a third point of very great interest , which is involved in Table XXXIV .
Here we have the coefficients of variation given for some twenty-two series .
Now this coefficient seems to me the only satisfactory comparative measure we can find present of variability .
A variation of two in the petals of a buttercup is far more significant than one of two in the florets of an ox-eyed daisy ; the measures of the absolute val.iations as given by the standard deviations seem to me of no use when we are comparing diffel.ent characters in different species .
In default of the suggestion of any better standard , all we can do is to get rid of absolute size or number by using the percentage variation of the character as indicated in the coefficients of variation .
Now table gives a fairly continuous series from up to The mean value of 19 is in very good accord with the results I have obtained variation in a much wider series for the vegetable kingdom .
Now I think it will be admitted : that this variation is based on results for a wide number of species ; that this variation covers a considerable yariety of characters ; that it is roughly continuous in value and 40 .
But if we examine the fourth column in this table , which gives the place of the homotypic tion , we find absolutely no relationship between the intensity the holnotyposis and of the variability .
The homotypic order will not compare in any way with the variation ordel .
The mean yariability of the first eleven sel.ies is , and of last eleven series , very sensible deviations from the leaIl 1 of the whole twenty-two series .
But the mean homotypic correlation of first eleyen series is 4559 , and of the last eleven series 4581 , neither of which If we excluded the hartstongue as largely influenced by environment , we should find broom heading the list , one of my most series , oth as to similarity of individual environment and as to of equoncy .
We should still Ilcludo ttion might to HOMOTYPOSIS IN THE yEGETABLE KINGDOM .
exhibits any sensible difference from the mean of all twenty-two series .
We are compelled therefore to conclude that there is no relationship between the variability of a species with regard to any organ , and the homotypic correlation of such organs .
It would be quite impossible to assert that as the val.iability of a species decreased , its homotyposis increased .
But according to the results reached in this paper heredity is a result of homotyposis .
Hence there seems , so far as our researches go , no ground TABLE XXXIV.\mdash ; General Results for Variability .
for asserting that increased int , ensity of heredity means decreased intensity of tion and vice versed .
The general impression upon my mind by our ments is the approximate constancy of the pure homotypic factor , 01 .
at any ts variation within a fairly small ange , and consequently the te constancy within a like small range of the hereditary factor for all species and characters .
VOL. CXCVII .
\mdash ; A. On the other hand , diversity in the variability of different species .
I do not wish to be matic on this point , however , but I would definitely assert that so far no relationship between variability and homotyposis is discoverable .
Further , it does not seem to me that looking down the order of variability we can assert that our table in any way corresponds to the order of evolutionary descent .
It does not seem to me that the species towards the top of it are the more primitive and those towards the bottom the more complex .
Indeed two series of the same species may stand at very different points in the table .
I am accordingly forced to the conclusion that my table gives no support to the view that variability in the vegetable kingdom is greatest for the more primitive and simpler organisms .
Nor , again examining Table XXXII .
, can we conclude that homotyposis is least intense for such organisms , the second half of our table cannot be said to contain more of such organisms than the first .
Shortly , I see no relation between the position of an organism in the evolutionary descent and the intensity of either its variability or its homotyposis .
Hence it follows that , if homotyposis be the source of heredity , the intensity of the latter cannot be related to either variability or position in evolutionary descent .
Mr. ADAM SEDGWICK may be right in his that in the early stages of evolution yariability is large and inheritance small , and that with development this relation is changed , variability decreasing with an increasing intensity of heredity .
My statistics only cover a small portion of one kingdom .
But so far as I am aware they represent the only material so far published on which a definite judgment can be founded\mdash ; where by definite judgment , I mean one based on quantitative facts , and not drawn from a general estimate of nature which makes no statement whatever of the intensity of either yariability or heredity for any single character in any one species of living things .
The data I have given seem entirely opposed to Mr. SEDGWICK'S view , but I very fully ealise that they are far from exhaustive .
They indicate , however , what I hold to be the only valid method of approaching any problem in variation or heredity , i.e. , ious statistical collection of actual facts .
My statistics may be insufficient , I would heartily welcome additions to them ; or they may be dealt with fallaciously .
Here is the material ; let others add to and , if need be , modify the conclusions .
Only let us follow the method so clearly indicated by DARWIN himself in his Across and Self-fertilisation of Plants ; ' let us cease to propound hypotheses illustrating them by isolated facts or vague generalities ; there are innumerable species at hand in Nature ready for us to measure and count .
Sine numero nihil dum est , should now-a-days be the motto of every naturalist who desires to propound novel hypotheses with regard to variation or heredity .
It is equally valid for those who merely seek either to establish more firmly or to modify the old .
( 26 .
) I would sum up the results of this memoir in the following conclusions : Addrcss to of the British tion , 1899 .
HOMOTYPOSIS IN THE yEGETABLE KINGDOM .
When an individual of any species produces a number of undifferentiated like organs ( homotypes ) these are not exactly alike when measured with regard to any character .
They exhibit a certain degree of variation combined with a certain degree of likeness ( homotypic correlation ) .
( ii .
) The homotypic correlation as tested for twenty-two series of homotypes in a variety of species in the vegetable kingdom , has a mean value between 4 and 5 .
Its exact determination is rendered difficult by a number of conflicting factors ; but it seems very possible that the actual value of pure homotypic correlation , i.e. , the correlation due to the individuality of the plant as apart from the influence of environment , \amp ; c. , differs but little from the mean value stated above .
( iii .
) If a character occurs in a series of homotypes , it will be found to have in the individual a variability 80 to 90 per cent. of the variability of this character in the species at large .
It is thus seen that variability is a primary factor of living forms , and is not in any way dependent on sexual reproduction .
( iy .
) With a certain hypothesis as to cross homotypic correlation , i. e. , the correlation of two different characters in homotypes , namely , that it is the product of the direct homotypic and of the organic correlations , it would follow that the mean fraternal correlation would equal the mean homotypic correlation .
The mean of twenty-two homotypic series is found to be sensibly identical with the mean of nineteen fraternal series .
A direct investigation of the value of the cross homotypic correlation is only made for two cases , that of the mushroom gill and that of the ivy leaf The cross homotypic correlation is not found to differ very widely from the product of the direct homotypic and the organic correlations in these cases , but the results are not close enough to be conclusive .
It would seem , if be true , that heredity is only a phase of the principle of homotyposis ( ii .
) , and that the numerical value of its constants may be found from that principle .
No relation whatever could be found between the intensity of variability and that of homotyposis in the twenty-two series dealt with .
Nor was any relation to be observed between the relative simplicity of the organism and the intensity either of its variability or its homotypic correlation .
Regarding heredity as a case of homotyposis , there seems no reason to suppose , as it has been suggested , that variability has decreased and heredity increased in the course of evolution .
On the contrary there seems some ground for supposing that homotyposis ( and therefore heredity ) is a primary factor of living forms , a condition for the evolution of life by natural selection , and not a product of such selection .
If the mushroom , the poppy , and the beech show approximately equal homotyposis , it seems well nigh impossib to consider it as a factor of life , increasing with advancing evolution .
PROFESSOR K. ARSON AND OTHERS ON On the Homotyposis of the Pods of Leguminous I have already indicated that I had initially grave doubts as to whether a true measurement of homotyposis could be obtained by counting the ripe seeds in the pods of leguminous plants ; but that a passage in a work of DARWIN'S led me to modify my opinion , and when finishing my memoir , I found time to incorporate in it the broom series from Danby Dale which seemed to confirm views .
Broom is cross-fertilised and much visited by insects , and accordingly the statement made on p. 334 of this memoir appears to have full validity .
As the matter seemed of some interest , I suggested to Miss C. D. a fuller study of leguminous plants from this standpoint .
Her results have reached me in time to be reduced by Dr. LEE and to be incorporated in this Appendix .
I have myselfworked series for the common bean and the tare vetch , so that if these be included we have now a fairly extensive range of material from which to draw conclusions .
I will first give the data and state the manner in which it has been reduced , and then consider the results which flow from its consideration .
The following is a list of the species dealt with ; I have remarked on some of the difficulties arising in the course of our work .
Cytisus ) .
Broom from Danby Dale ( see p. 334 ) .
\mdash ; It was found difficult to count the total number of ovules , and only the ripened seeds were dealt with .
The plant is cross-fertilised .
The pods were fairly free from insects .
( ii .
) Lotus Corniculatus .
Bird's-foot trefoil from the neighbourhood of Lyme Regis , Dorsetshire.\mdash ; The aborted seeds could not be counted , and the pods were largely frequented by grubs , so that it was occasionally doubtful how far ripe seeds had been destroyed .
The number of pods on each plant in this first series were too few too allow those with grubs in them to be neglected .
* ( iii .
) Lotns .\mdash ; A second series from the neighbourhood of Lyme Regis .
This series consisted of more vigorous plants from a slightly different environment , so that fourteen to twenty pods could be gathered from each , and thus difficulties as to the possible destruction of ripe seeds by grubs avoided .
It would appear that the effect of rejecting such pods has raised the mean number of seeds , but at the same time artificially reduced the homotyposis .
The plant is cross-fertilised .
( DARWIN : ' Cross and Self-Fertilisation of Plants , ' p. 361 .
) .
Sweet Pea.\mdash ; This was obtained from a row in a nursery garden at Lyme Regis .
Both the fertilised and aborted seeds were counted , so that the total number of ovules could be found .
The plant is said to be cross-fertilised in * As an illustration of this sort of difficulty I may note that I found something like 90 per cent. of the pods of gorsc in the Danby district so full of insect life that nothing could be done in regard to counting their sccds .
HOMOTYPOSIS IN THE VEGETABLE KINGDOM .
countries where it is indigenous , but DARWIN ( ' Cross and Self-Fertilisation of Plants , ' 3rd edition , p. 155 , et seq. ) strongly believes it to be wholly self-fertilised in England .
Lathyrus Odorafus.\mdash ; A shorter series obtained from a second nursery garden at Lyme Regis .
Sylvestris .
Everlasting pea from the sea coast not very far from Lyme Regis .
\mdash ; Both the ripe and the aborted seeds were counted .
The plant appears to be cross-fertilised .
Professor F. O. OLIVER tells me that he has watched bees effectiyely visiting the cultivated everlasting pea .
( vii .
) Faba .
Common Bean.\mdash ; I took 100 plants , each having at least ten pods , from a field in Danby Dale , in which a mixture of oats , the common pea , the common bean , and three or four vetches* ( described by the owner as tares ) , was growing as food for cattle .
The perfect beans only were counted , but the plants were a poor crop , many pods being stunted in their growth , and it was not always easy to determine whether the seeds had not been fertilised or there had been failure owing to want of nutrition .
The common bean is both cross and self-fertilised .
I have to thank Miss J. SHARPE and Miss E. CYRIAX for aid in the work on these beans .
( viii .
) Vicia .
\mdash ; I found twenty-eight plants of the tare vetch growing on a strip of uncut grass along a cut cornfield at Botton , Danby Dale .
I could ffid no more plants in the neighbourhood .
There was plenty of the common vetch on the same strip , but the plants had rarely more than two to three pods on them .
I was not able to get ten pods from each tare vetch , thirteen plants had fewer , and seven plants more .
The series being short , the probable errors are high , but it seemed worth while to include the data .
I endeavoured to count both ripe and aborted seeds , but here , as in one or two of previous cases , I much doubt whether we have succeeded in counting all the ovules .
In some of the green pods the number of ovules seemed to be considerably larger than in the fully ripe pods , and I think it possible that the non-fertilised seeds shrink till they are quite unnoticeable even by a cautious observer .
Again , it is by no means certain here , as in other cases , that all the seeds reckoned as aborted are really non-fertilised .
It is quite possible that in some cases fertilised seeds have dwindled for want of nutriment till they appear aborted .
Vicia is either cross or self-fertilised ( DARWIN : Across and Self-Fertilisation of Plants , ' p. 367 ) .
Whether this vetch , or indeed the common bean , were in our case wholly cross-fertilised or in part self-fertilised , I do not see that we have any means of settling .
Table L. gives the frequency distributions of the various series .
Now in these series we must bear in mind that we are only definitely certain of one thing , the number *Neither the pea nor the vetches offered enongh pods per plant for an investigation of their homotyposis to be of value .
of fully ripened seeds\mdash ; and this only when the pods were not infested with grubs .
We did the best we could to count the aborted seeds , but we cannot be certain that all those counted as aborted were non-fertilised , or that we succeeded in counting all those which had to microscopic proportions .
The number of ovules is simply the sum of the ripe and the aborted seeds counted in each pod , and this again must be somewhat doubtful .
The reader will bear in mind that we proceeded with care , but that we thin it right , if anything , to rather over- than under-emphasise possible sources of error .
Yet allowing for such sources of error we cannot , on examining the results given in Table LI .
, allow that Broom is in the least representative of the degree of homotyposis to be found in the pods of leguminous plants .
Whether we consider the ovules , the ripe , or the aborted seeds , our results are sensibly below that for broom , and the mean of the whole sixteen series gives us a value about one-half that of the homotypic correlation based upon characters not depending on fertilisation .
If we deal with averages , it would certainly seem that in the results flowing from fertilisation , we have reduced the intensity of the individuality to about half its previous value .
To this extent only does the individual constitution appear influential in the number of in the pod , the remainder of the homotypic intensity seems to have disappeared under random influences having nothing to do with the individuality of the plant .
This is perhaps what we might expect in the case of seeds in crossfertilised plants , where the fertilisation may depend on the chance or not of insect visitation and the effectiveness or not of the pollen brought on such occasions .
We should have to assert that the bird's-foot trefoil and the everlasting pea , whose average for homotypic correlation in the case of ripe seeds is about to 22 , lose half the intensity of their individuality through the random nature of the chances of the cross-fertilisation .
But although this might be fairly satisfactory for these cases , what are we to say for the species which are self-fertilised absolutely or self-fertilised in default of cross-fertilisation ?
We might have expected a high degree of homotyposis in the field bean or the tare vetch , where failing cross-fertilisation we are told there will be self-fertilisation .
We find on the contrary , however , in these results some of the lowest homotypic correlations of the whole series .
The sweet pea also in its two series presents some very remarkable results .
If the sweet pea be entirely selffertilised then we should expect the homotyposis of both ovules and ripe seeds to be fully up to the average .
In neither case is such a result reached , although in one series we have a value for the ovules higher than that obtained for anything except the broom .
The noteworthy fact , however , is that the results for the two series of sweet peas differ so widely in character !
In the first series it is the abortion which is most individual in character ; in the second series it is the ovules .
In the first series the ripe seeds have far less individuality than the aborted seeds , in fact , the smallest homotyposis I have yet observed ; in the second series the ripe seeds have a value rising to 2 , which is higher than that of the aborted seeds .
Ash HOMOTYPOSIS IN THE KINGDOM .
trees from Monmouthshire , Buckinghamshire , and Dorsetshire , gave results in good agreement ; but sweet peas from two different nursery gardens in the same district give strikingly divergent homotypic correlations !
According to DARWIN ( ' The ffects of Cross and Self-Fertilisation , ' edition , p. 153 ) the sweet pea " " in this country seelns invariably to fertilise itself\ldquo ; He bases this statement on ( a ) the difficulty of access to bees and other insects , ( b ) the fact that the varieties are habitually grown by seed growers close together , and yet the colours do not blend , ( c ) the experience that when the varieties are artificially crossed the colours do change .
Professor DXLPHINO , of lorence , in a letter to DARWIN , writes that it is the fixed opinion of the Italian gardeners that the varieties do cross ; but in Italy other insects may of course be available .
Now it is most remarkable that in Series I. , whether the sweet pea be crossed or self-fertilised , number of seeds should be hardly individual at all .
It would indicate that , with the particular environment of this series , the chance even of self-fertilisation depends upon extraneous causes .
The comparatively high individuality in the tendency to abort may mark some peculiarity of this variety ; there may be a distinct inheritance of sterile tendencies to be kept separate from an inheritance of fertility .
But it is probably idle to guess at explanations of such crepancies .
I have gone carefully into the differences of environment in the two series which are interesting , although they do not obyiously provide any key to the mystery .
The first series were thickly planted in a long row , and the seed pods were gathered unripe .
It was more difficult to separate each and to be quite certain that some of the seeds would not have had to be reckoned as aborted in the fully ripened pods .
The second series consisted of withered plants , the pods bein almost all completely ripe ; the pens were planted in separate groups a few feet apart .
There was no difficulty in ascertaining the individuality of the plant nor as to the number of ripe or seeds .
About forty hives of bees were kept in the immediate neighbourhood of this second series ; the first series were at the other end of Lyme Regis , and removed from hives .
Both series had had their flowers freely cut , I might been prepared to attribute the low value of the homotyposis in the sweet pea to this latter cause , but then the common bean and the tare vetch had not been subjected to any similar process , and they give on the whole much lower values .
The ovules in either self-fertilised or cross-fertilised plants ought indeed to give a of homotyposis , but , for reasons already stated , it is not certain that our procedure in counting the perfect and the aborted seeds in the ripened pods has led to ajust estimate of them .
I am inclined on the whole to attempt no explanations for the anomalies observed in these seed investigations .
The influences of self- and cross* I mention this , as I noticed in Qhire fairly frequent but pparcntly i ] visits of the ve bees to sweet peas .
It is possible that their attempts may ssist the tbion , just sh .
or wind has been observed to do in species .
VOL. cxcvn.\mdash ; A. 3 fertilisation on homotyposis deserve still fuller and more direct investigatio I feel we know little as to the influence of external causes even on the completeness or incompleteness of self-fertilisation .
So far as homotyposis in the pods of leguminous plants is concerned , I would draw conclusions based solely on averages , and state that : \mdash ; The homotyposis in the case of either ripe or aborted seeds in the pods plants seems weakened to one-half the average value it has in the case of characters not depending on fertilisation .
Thus DARWIN'S view , that differences in the number of ripe seeds depend upon the constitution of the plants , seems to be only partially true .
Extraneous causes about which we very clear appear to be generally influential .
( ii .
) The extraneous causes which act in a random manner on the homotypic correlation seem to affect both self-fertilised and cross-fertilised plants , and this both with regard to ripe and to aborted seeds .
( iii .
) The order of intensity of homotypic correlation is oyules , ripened seeds , and aborted seeds .
The reduction of homotypic correlation in the case of the fertility of pods may be profitably compared with the like reduction which we find in the case of the coefficients of inheritance of fertility and fecundity in man and the horse .
Other points worth noting , I think , in our results are those of Table LII .
, which gives the correlation between ovules and ripe and aborted seeds .
As we might anticipate , the more ripe the fewer aborted seeds , and vice versed .
The correlation is negative , high for the .everlasting pea , remarkably low for the tare vetch .
It is interesting to see that the correlation between the ovules and either ripe or aborted seeds is not very high .
A large number of ovules not necessarily connoting either a very large number of ripe or of aborted seeds .
Tn fact , in the case of the everlasting pea , the number of ovules has very small influence indeed on the number of seeds which ripen .
In the tare vetch only is the relationship more marked .
Professor .
O. OLIVER tells me that in certain cases evolution appears to be tending in the direction of the pod containing one ripe seed only .
It seems , therefore , that such relations as are indicated in Table LII .
( and others of a like kind , for further observations ought certainly to be made ) may be useful in indicating the degree of fixity between the number of ovules and the number of ripe or aborted seed which are ultimately to be found in the seed vessel .
While the ovules have the least and the aborted seeds the greatest yariability , as measured by the coefficient of variation , the ovules have the most and the aborted seeds the least homotypic correlation .
This might at first sight appear to be opposed to the view expressed on p. 363 , that there is no relationship between the intensities * It seems to me that investigations of this kind ought to be carried out by those who have , what I unfortunately have not , the needful land for experimental investigations , ( Phil. Trans , vol. 192 , p. 277 et IN THE VEGETABLE KIN of variation and of homotyposis .
But it is really not so , .
if we proceed as on that page to divide up our series into two groups in order of their variability , we find that the first eight give an ayerage coefficient of variation of , and have an average homotypic correlation of 1941 ; the last eight have an average coefficient of variation of 5147\mdash ; double and more that of the first eight\mdash ; while their homotypic elation is 2207 .
Thus , so far from the homotyposis being reduced by greater bility , it actually appears to be slightly increased .
Lastly , we may notice ( Table ) that the variation in the individual amounts to upwards of 97 per cent. of that in the race , or we have another strong fact to aid in demolishin the theory that variability is a result of sexual reproduction .
Note in particular the 97 per cent. in the case of the ovules whose number is mined before fertilisation !
Tables XXXV.-XLIX .
give the data from which the constants of Tables L. have been calculated .
HOMOTYPoeIS IN THE VEGETABLE KINGDOM .
\amp ; om Monmouthshire , Buckinghamshire , and Dorsetshire , gave results in good ; but sweet peas from two different nursery gardens in the same district strikingly divergent homotypic correlations !
ccording to DARWIN ( ' The Effects of Cross and Self-Fertilisation , ' 3rd edition , 153 ) the sweet pea " " in this country seems invariably to fertilise itself\ldquo ; He bases statement on ( a ) the difficulty of access to bees and other insects , ( b ) the fact that varieties are habitually grown by seed growers close together , and yet the colours not blend , ) the experience that when the varieties are artificially crossed the do change .
Professor DELPHINO , of Florence , in a letter to DARWIN , writes it is the fixed opinion of the Italian gardeners that the varieties do cross ; but in Italy other insects may of course be available .
Now it is most remarkable that in , whether the sweet pea be crossed or self-fertilised , the number of ripe seeds be hardly individual at all .
It would indicate that , with the particular environment of this series , the chance of self-fertilisation depends upon extraneous causes .
The comparatively high individuality in the tendency to abort may mark some peculiarity of this variety ; there may be a distinct inheritance of sterile tendencies to be kept separate from an inheritance of fertility .
But it is probably idle to guess at explanations of such discrepancies .
I have gone carefully into the differences of environment in the two series which are interesting , although they do not obviously provide any key to the mystery .
The first series were thickly planted in a long row , and the seed pods were gathered unripe .
It was more difficult to separate each plant nd to be quite certain that.some of the seeds would not have had to be reckoned as aborted in the fully ripened pods .
The second series consisted of withered plants , the pods being almost all completely ripe ; the peas were planted in separate groups a few feet apart .
There was no difficulty in ascertaining the individuality of the plant nor as to the number of ripe or aborted seeds .
About forty hives of bees were kept in the immediate neighbourhood of this second series ; the first series were at the other end of Lyme Regis , and removed from hives .
Both series had had their flowers freely I might have been prepared to attribute the low value of the homotyposis in the sweet pea to this latter cause , but then the common bean and the tare vetch had not been subjected to any similar process , and they give on the whole much lower values .
The ovules in either self-fertihsed or cross-fertilised plants ought indeed to give a fair measure of homotyposis , but , for reas5ns already stated , it is not certain that our procedure in counting the perfect and the aborted seeds in the ripened pods has led to ajust estimate of them .
' I am inclined on the whole to attempt no explanations for the anomalies observed in these seed investigations .
The influences of self- and cross* I menbon this , as I noticul in hire fairly frequent but apparently ineffectual visits of the hive bees to sweet peas .
It is possible that their mpts may assist the self-fertilisation , just as shaking or wind has been observed to do in other species .
VOL. CXCVIL\mdash ; A. S

