K. Pearson 
161 
Appendix Ia. 
[Ani/ Teacher willing to give assistance in these observations — an assistance which will be duly 
acknowledged in the final pvMication of residts—is requested to communicate with Professor 
Karl Pearson, F.R.S., University College, London?\ 
GENERAL DIRECTIONS FOR FILLING UP DATA PAPERS OF 
COLLATERAL HEREDITY*. 
1. The object of this investigation is two-fold : 
(i) To ascertain the degree of resemblance, mental and physical, between children of 
the same parents. 
(ii) To discover, if possible, whether there is any relationship between the external 
shape of the head and a teacher's estimate of the general grade of ability of the 
pupil. 
Co-operators are warned ab initio that no inferences whatever can be drawn from individual 
instances or from a small series of measurements. The numerical quantities to be determined 
are small, and it is only when large masses of observations have been collected from many 
quarters and have been reduced that reliable inferences can be drawn. 
2. The measurements and estimates are to be made on : 
(i) Pairs of brothers (white data paper). 
(ii) Pairs of sisters (pink data paper). 
(iii) Pairs of brothers and sisters (blue data paper). 
Care must be taken that the right coloured data paper is selected. 
The names of the measured are only required in case there should be need for the verification 
of any entry, and they will be treated as strictly confidential. Initials, in fact, may be used 
where it seems desirable, if the observer keeps a key to them for the purpose of reference should 
reference be required. 
The observer should have known well both members of the jjair measured for at least six 
months, and, if possible, for a much longer period. The classification is purposely made rather 
wide and indefinite in order that there may be less hesitation in classifying. What is needed is 
the general impression of a teacher who has carefully observed his or her pupils. 
For both physique and ability it is very desirable that the observer should consult, where it is 
possible, one or more colleagues before filling up the data paper. 
To give some confidence in the scales adopted, I may remark that in response to my appeal 
in the Journal of Ediication, I received details of some 150 boys and girls tested for ability by 
three observers independently (language, science, and mathematical teachers) and belonging to 
half-a-dozen different schools. The agreement in classification was complete in more than 
80 per cent, of cases, and only differed by as much as two classes in about 5 per cent, of cases t. 
This degree of accordance is sufficient for the present statistical purposes. 
3. I. Physique. In making the record, attention should be paid not only to appearance, 
energy, and athletic qualities, but to irregularity of attendances owing to ill-health, frequency of 
visits to school-infirmary, etc. 
* The quantitative laws of heredity, such as we have reached at present, do not apply to individual 
cases, but only to the averages of large numbers. It is important to insist on this, because more than 
one of my helpers on hearing the results of a particular research has seemed disappointed, remarking 
that the law does not hold for the family X or the brothers Y. 
t Even this amount of divergence would probably have disappeared after a consultation with regard 
to the individuals classified. 
Biometrika iii 
21 
