INTRODUCTION.

At the time of the first publication of Mr Galton's Hereditary Genius, in 1869, the belief in the hereditary nature of inborn natural ability was held by very few; but so great has been the influence of that and other works that at the present time it would be almost impossible to find an educated person to dispute it, and the inheritance of psychical characters ranks with motor-cars and the morphia habit among the ingredients in the composition of the contemporary novel.

With the diffusion of an idea it becomes progressively more vague and indefinite, so that an attempt to give it crispness of outline becomes all the more desirable, and for this purpose one must substitute, for a description that is merely qualitative, one that is also quantitative, and, having recognized the pressure and direction of a force, one must endeavour to measure it.

Such a description of the inheritance of the mental and moral qualities has been given by Professor Pearson in his Huxley Lecture before the Anthropological Institute, and it has been our endeavour to confirm and supplement his results with others obtained from different material treated at treated at the outset in a fundamentally

different way.

The essential difference between his results and ours is that the former are based on the estimates of the school teachers of the ability of pairs of brothers, or sisters, or brothers and sisters, while we have used the results of the examinations for the B.A. degree at Oxford, the position in the school of boys in the school at definite times in their life or in their school career. This has enabled us with regard to the Oxford material to measure the average resemblance between father and son, as well as that between brother and brother, and we shall endeavour to show that, after certain allowances have been made, our results are substantially in agreement, not only with those of Professor Pearson, which are referred to above, but also with other work on more easily measurable characters.

Perhaps it would be as well to define, as exactly and as simply as we can, the meaning we attach to the phrase "measure the average resemblance between father and son." We know that the intelligence of the fathers varies very much and so also does that of the sons; we know in a general sort of way that the sons of those fathers who are above the average of fathers, are themselves on the whole more intelligent than the average of sons; if we take a group of fathers,

G. M.

1

each of whose intelligence is the same definite amount, which we will call A*, above the average, their sons will form a group which varies considerably, but whose average intelligence is a certain definite amount, which we will call B*, above that of all the sons of all the fathers.

We presume, and there is considerable reason for the presumption, that whatever be the magnitude of A the relation of B to A, that is to say the value of the fraction B is, within certain limits, the same, and we want to find out exactly what it is. When we have found it we shall have measured the average resemblance between father and son, for if the resemblance be complete and perfect then the value of this fraction would be 1, for B would be equal to A, that is to say that the sons of any particular group of fathers of given intelligence will differ from the average of all the sons by exactly the same amount as the intelligence of their fathers differed from the average intelligence of all the fathers. If there be no resemblance the value of the fraction is 0, and every intermediate value between 1 and 0 expresses a different degree of resemblance.

A statistical statement of the effect of inheritance on the mass such as is attempted here is quite independent of what view is taken of the truth of Mendel's laws and whether they be universal in their application or not; this fact has been frequently insisted upon, yet it seems desirable to do so once again. Even if we believe in the universal applicability of these laws-and it cannot be claimed that this has as yet been proved-they would tell us nothing concerning the inheritance of inborn ability until such a time as some general law of dominance has been discovered and some better definition of a unit character evolved than that which is at present accepted-namely a character that is inherited according to these laws. Furthermore, granting that this time had come, Mendel still would tell one nothing unless in each case both parents were known, and through a necessary limitation of the material this condition does not obtain. Enough has been said to show to any but the extremists of the Mendelian school that this work, if properly executed, is of value, and that if this is not the case the blame must be attached to the authors personally, and not to the methods which they employ nor to the school to which they belong.

*For the sake of simplicity we have omitted to state in the text that A is measured in terms of the standard deviation of fathers, and B in terms of the standard deviation of the sons.

PART I.

THE EVIDENCE OF THE OXFORD CLASS LISTS.

A. Sources of Material.

THE standard biographical works concerning past and present members of Oxford University are Foster's Alumni Oxonienses and its continuation Oxford Men and their Colleges. The former is divided into two parts, each consisting of four octavo volumes an earlier one covering the years 1500-1714, and a later one, which deals with those who entered the University between 1715 and 1886, while the continuation brings the information up to 1892. Since the Oxford honours examinations on which our tables are largely based were only instituted in 1800, only a portion of the later part could be employed, but it will be seen that in spite of this an ample amount of material has been obtained.

The biographical notice of each member of the University gives, among other information, the name of his father and the name of the college to which he belonged; it tells whether or no he took the B.A. degree, and if so at what date. Thus from this we were enabled to find the relationship between a large number of pairs of fathers and sons, and of a large number of families of brothers, while the remaining information which we required, namely the class obtained in the case of an honours degree, was supplied by the Oxford Historical Register. The last-named work contains among other useful features an alphabetical register of all those who obtained honours or distinctions at the University between 1220 and 1900; after each name is given the college to which he belonged, which is useful for purposes of identification, and a dated list of the honours and distinctions obtained. Thus the labour of finding out which class, if any, was obtained by each of the several thousand persons included in our tables was reduced to a minimum. The history of the honours examinations at Oxford, which may be found summarized on p. 191 of the Historical Register, stated still more briefly, is as follows: From 1800-1806 the honours examination was conducted separately to the pass examination, the candidates were examined both in classics and mathematics, and there were two classes of honours, but during those years, either very few people entered for honours, or the standard was exceedingly high, for only 14 men appear in either the first or second class. From 1807-1830 alternative subjects were introduced, namely Literae Humaniores or classics, and Disciplinae Mathematicae et Physicae or mathematics; all candidates, whether

for honours or no, were examined together, and there were in 1807-8 two classes of honours, which were increased to three in 1809. In 1830 an extra class was added and those who were not candidates for honours were examined separately from those who were, but permission was given to the examiners to include in the fourth class honours list those candidates for the pass degree whom they thought worthy of it. This enactment came to an end in 1865, when the honours. examination and pass examination were allotted to different bodies of examiners. From 1830 onwards the four classes of honours have remained the same, although the number of subjects in which the examination is held has been greatly increased.

Up to the year 1834 the members of New College were exempt from the University examinations, a privilege which shut out that college from the “rapid improvement in industry and intellectual vitality which that measure brought with it for the best of Oxford colleges." At that date it was voluntarily renounced by the college, but up to then New College men can hardly be considered to have been members of the University as far as examinations are concerned and are therefore not included in the tables which we have made.

B. The Resemblance between Father and Son.

The first step in the process of tabulation was to divide the men into three classes according to the date at which they took, or should have taken their degrees, the limits of the three groups being as follows: 1800-30, 1830-60, 1860-92. We will consider the latest of these first, namely the group 1860-92, and deal only with those men whose fathers were educated at Oxford. They were found to consist of 2459 persons, of whom 149 had been placed in the first class, 329 in the second, 377 in the third, 190 in the fourth; of whom 868 had taken merely a pass degree, and 546 had failed for one reason or another to take any degree at all. Each of these six classes was sorted again according to the degree taken by the father; of the 149 first class men, 27 had fathers in the first class, 27 in the second class, 14 in the third class, 13 in the fourth, and 53 with pass degrees and 15 without degrees; of the 329 second class men, 52 had fathers in the first class, 54 in the second class, 33 in the third class, 30 in the fourth class, 138 with pass degrees, and 22 with no degrees; of the 377 third class men, 47 had fathers in the first class, 64 in the second, 47 in the third, 42 in the fourth, 157 with pass degrees, and 20 with no degrees; of the 190 fourth class men, 20 had fathers in the first class, 27 in the second class, 22 in the third class, 17 in the fourth, 91 with pass degrees, and 13 without degrees; of the 868 men who took pass degrees, 41 had fathers in the first class, 79 in the second, 95 in the third, 87 in the fourth, 479 with pass degrees, and 87 with

no degrees; of the 546 who took no degrees, 31 had fathers in the first class, 39 in the second class, 46 in the third class, 52 in the fourth, 277 with pass degrees, and 101 with no degrees. These results will be found stated in a less distressing way in Table I A., but at the risk of seeming, or possibly it might be more accurate to say, in spite of the certainty of being insufferably tedious, we have repeated them in words here, so that those people who are quite unfamiliar with this class of table should have no difficulty of apprehending its meaning.

Sons with first class honours

29.4

24.7

DIAGRAM I. (date of degree of sons, 1860-92).

The heights of the vertical lines show in what percentage of cases the fathers have taken either first class or second class honours. The diagram is intended to show that the percentage of fathers who obtained this degree of distinction diminishes with some regularity, as one passes downwards from the sons with first class honours to those with no degrees.

Even without any mathematical treatment we can learn a great deal concerning heredity from it. Thus it will be seen from Table IB. that of the fathers of the first class men 362 per cent. obtained either a first or a second class themselves, and thus were on the whole slightly superior to those of the second class men of whom only 32.2 per cent. reached this standard; a stage lower is reached in the fathers of the third class men, among whom the percentage of first and second