306 REPORTS ON THE STATE OF SCIENCE, ETC. 
irrespective of the length of the operations that are being tested, and which 
constituted the ‘ trial.’ 
The figures also showed that, so far as the first five trials that constituted 
these tests were concerned, one ‘ trial’ was about as reliable as another. 
A similar result was observed during the much longer practice period, for 
it was found that the reliability of the ‘test’ was much the same from 
one day to another, irrespective of the stage of the subjects’ practice. 
Generally, the tests measured ‘ initial’ ability to about the same degree 
of accuracy as they predicted ability after practice. 
Little difference in reliability was observed between the adults and 
elementary school subjects. 
Incentives—The correlation between the scores made at the various 
tests, and estimated ‘incentive,’ proved to be negligibly small. The 
results which we have now to examine can, therefore, hardly be explained 
on the basis of differences in incentive. 
Il. THe SpeciaL ABILITIES (OR GROUP-FACTORS). 
The Mechanical Factor.—Having secured reliable measures of ability, our 
next step was to determine the intercorrelations of all the tests in the 
case of each group of subjects tested (i.e. the ‘ adult ’ group, and the several 
elementary school groups). It was at once evident from these correlations 
that the data tended to fall into three groups, viz.: (i) a ‘mechanical’ group 
consisting of the mechanical aptitude tests and the mechanical assembling 
tests ; (ii) a routine (or ‘manual’) group composed of the routine assembling 
and stripping tests, and the simple manual tests, and (iii) a general 
intelligence group consisting of the tests and estimates of intelligence, and 
general school subjects. 
The next step was to determine, by Spearman’s method of tetrad- 
differences, how far these observed differences in the correlation coefficients 
are due to chance, or to differences in the degree of correlation which all 
of the other tests showed with the intelligence group. The application of 
this criterion indicated that although the general positive correlation running 
throughout the data could best be ascribed to a general factor common to 
all, there were also present group-factors, tending to produce a closer 
relationship between members of the same group than could be accounted 
for by this general factor. 
To determine more precisely the location and range of the group-factors, 
the influence of the general factor was next statistically eliminated and the 
tetrad-difference criterion was then applied to the resulting specific 
correlations. 
It is impossible to present here the numerous correlation tables examined 
in the course of the analysis. It must suffice to say that the following 
conclusions were clearly indicated’: 
(1) The specific intercorrelations of the ‘mechanical’ group: were 
best explicable by a single group-factor common to both the ‘ aptitude’ 
and the ‘ assembly’ tests. This seemed most reasonably identified with 
the mechanical factor (‘m’) which was disclosed in the aptitude type of 
test in a previous research, and whose presence was thus confirmed in the 
present research, and shown, for the first time, to be present in suitable 
tests of the mechanical assembling type. 
(2) The mechanical aptitude type (which, it will be remembered, involved 
no manual activity) were more highly saturated with the mechanical group- 
factor than were the assembling type, and therefore provide better measures 
of this special ability. 
