SECTIONAL TRANSACTIONS.—J. 441 
fluctuation is indicated by rx,y,, and rx,y, being greater than ry,y, and 
rx,y,; if no function fluctuation is present, these will all be equal. Alterna- 
tively, r(x, — x,) (Y,— Y,) Will be positive if function fluctuation is present, 
zero if itis absent. ‘This we may call the double test re-test criterion. Other 
methods of solving the problem are available for somewhat different kinds 
of test data. 
Mr. E. J. G. Braprorp.—The reliability of test measurements (10.45). 
The reliability of a test measurement is frequently estimated by comparing 
the results obtained from several applications of that test. Variation in 
the results may be due to experimental conditions (environmental), to 
different samples of the population tested, to different speeds at which the 
subjects of the experiment adapt themselves to the tests, to the different 
methods by which this adaptation is brought about. ‘The term reliability 
can be used in connection with each of these sources of variability. 
One test form or application may be reliable in one sense and a second 
test form or application in another sense; the correlation between them 
gives no indication as to the sense in which each is reliable. 
Considerable attention has been devoted to the first two sources of 
variability, and the deviations of the ‘ reliability coefficient ’ from unity have 
been ascribed to these sources ; whereas the remaining two sources of 
variation have probably more psychological significance. Differential 
cumulative adaptability (learning) and differential methods of adaptation 
(vicarious functioning) need to be considered in connection with the 
reliability of test results. Reliability is not entirely a statistical problem. 
Dr. W. STEPHENSON.—Some applications of the inverted factor technique 
(11.30). 
Dr. 5. J. F. Poitporr.—Problems in the field of output and oscillation (12.15). 
(a) Ambiguities in description: When there is fluctuation with a sound 
of minimal intensity (it being now heard, now lost) we tend either to say it 
is coming ‘ in and out’ of awareness, or that auditory ability is ‘ rising and 
falling.’ 
It is usual only to use the in-and-out conception when describing fluctua- 
tions with reversible perspective, one meaning being said to come in, as 
the other goes out. It is as customary to say that ability rises and falls 
when computation or a like task is being performed, although there is in- 
and-out switching in the sense that individual units of work come success- 
ively to the focus, or that we switch from one way of ‘ doing’ the task to 
another and back again. 
(6) Levels of rivalry : Whatever its nature, a task is performed in compe- 
tition with others ready to supplant it. Within it there is rivalry between 
its various aspects. Within any one aspect there is conflict between the 
ways in which it may be cognised. 
(c) Switching and Spearman factors, fatigue, etc.: Speed of switching 
in g-tests and p-tests means high ability and low inertia respectively. In 
reversible perspective it is normally held to indicate tiredness. In computa- 
tion and like tasks it is a necessary condition of high output. These various 
notions need harmonising. Incidentally, the narrower the field in which 
switching may take place, or the more intense the rivalry, the greater the 
degree of fatigue, and vice versa. 
