H. Waite 
91 
The divergence between the rj and r values, i.e. the absence of accurately linear 
regression is not, I believe, due to any inherent quality of the characters discussed. 
It results from selection according to age and ability peculiar to this school and to 
the arrangement of boys in their forms. It would probably disappear entirely 
were a number of schools with a far larger range of pupils investigated. For this 
reason it seems likely that the values of r, as found by the fourfold table, express 
the fundamental relations more closely than those found for r\. 
Note. As the masters are specialists they judge the capacity of the boys from 
somewhat different points of view. Under these circumstances it is worth while 
to compare the estimates which were made quite independently, and also without 
any knowledge, on the part of the masters, of the object for which they were 
required. 
In connection with the first of the above Reports the total number of pairs of 
judgments given was 1409 ; of these, 662 or 47 per cent, were in exact agreement ; 
684 or 48'5 per cent, differed by one place ; 62 or 4 - 4 per cent, differed by two 
places while in only one case was there a divergence amounting to three places. 
For the second Report 2018 pairs of judgments were given of which 976 or 48'4 
per cent, agreed exactly ; 944 or 46'7 per cent, differed by one place, 92 or 4"6 
per cent, by two places, and 6 by three places. Thus, in less than 5 per cent, 
of the whole of the 3427 pairs of judgments was there a difference of opinion 
amounting to more than one place in the scale adopted, although in all cases 
the judgments were based on ability shown in different subjects or groups of 
subjects. 
Again in estimating General Intelligence for the second Report, it happened 
that a master was called upon to give his opinion a second time on the same boy 
in 355 cases. The interval which elapsed was about nine months, and no record 
was kept by the masters themselves of the first set of records. In 202 cases the 
same opinions were given as on the previous occasion ; in 81 cases there was a rise, 
and in 68 a fall, of one place in the scale ; in only 4 cases was there a difference 
of two places, three of these being a rise. 
The above figures are an additional proof that the teacher's judgment of 
mental capacity is formed with careful discrimination and an intimate knowledge 
of his subject. 
The pairs of judgments discussed above may be examined by means of contin- 
gency tables. I have arranged judgments given by First Master in rows against 
those by Second Master in columns, the " First " and " Second " being fixed by the 
order of the masters in an arbitrary list; e.g., each case judged by a First Master 
as G and by a Second as D will be included in the [C, D] cell, but since in any 
pair of judgments either master may be considered the First Master, I have added 
rows to columns, thereby doubling the number of pairs and at the same time 
producing a symmetrical table. 
