440 SECTIONAL TRANSACTIONS.—J. 
series of tests of undoubted value in selection. This may be illustrated by 
examples of tests recently devised on this basis for a number of occupations, 
e.g. chemical process workers, machine operators, glass inspectors. Infor- 
mation on the technique of making job analyses and on the type of tests 
that prove of value is accumulating. 
Prof. T. NorTtH WHITEHEAD.—An observation of an American industrial 
group (3.30). 
This paper describes an analysis by some members of the Harvard 
School of Business Administration of an experiment initiated and conducted 
by the Western Electric Company at their Hawthorne Works, Chicago, 
Illinois. 
A small group of experienced assemblers engaged in repetitive work were 
minutely investigated for five years whilst pursuing their ordinary occupa- 
tion. ‘Their minute to minute outputs were automatically recorded and 
many other records were obtained, both numerical and qualitative. 
This industrial group simultaneously performed two distinct functions. 
(1) A technological, or economic function. 
(2) A sccial function, by which the group secured and maintained co- 
hesion, both within itself and in its external relations. The success of the 
former function was bound up with that of the latter. 
Marked variations in working rate were recorded ; but these were insensi- 
tive to changes in physical circumstance, temperature, relative humidity, 
hours of work, etc. ; they were mainly due to changes in social circumstance, 
both intra- and inter-group, chiefly the former. : 
Statistical methods are elaborated showing that in this instance inter- 
personal cohesion manifested itself in imitative, as distinct from merely 
related, behaviour; and that antagonisms resulted in unlike behaviour 
(negative correlation). Social disruption, as distinct from mere antagon- 
isms, resulted in independent (uncorrelated) behaviour. 
Mr. J. R. Jennincs.—The methods of industrial psychology applied to 
agriculture (4.15). 
Tuesday, September 10. 
Joust Discussion with Section G (Engineering) on The applications of 
science to the control of road traffic (Section G room) (10.0). See 
under Section G, p. 417. 
Dr. R. H. TuHouLess.—The distinction between test unreliability and fluc- 
tuations of mental functions (10.0). 
The results of a mental test applied twice to the same group of subjects 
do not generally give complete correlation. ‘This failure in complete 
correlation may be due to either or both of two causes: (a) the test may be 
unreliable, i.e. when applied successively as a measuring instrument for the 
same unvarying quantity, it may not give self-consistent results; (b) the 
function which is measured by the test in any one subject may vary quanti- 
tatively from day to day. It is important to distinguish between these two 
cases. ‘Test unreliability is no doubt always present, but it is of importance 
to know whether fluctuation of the function measured is also present and 
to be able to estimate its amount. If two tests (X and Y) of the function 
in question are applied simultaneously on two successive occasions, function 
