APPRENTICE TOOLMAKERS AND MACHINISTS 125 
was asked to rank the men in the order of their ability 
to learn the prescribed work. At the same time, he was 
asked also to rank his men according to the opinion he had 
formed of them when they first came into the shop. The 
Method by which these ranks were obtained is probably 
Worth describing for the aid which it may be to other 
experimenters. 
It had been found that, usually, when a foreman is 
asked to rank the men or women under him, he is unable 
to rank them in any but the crudest form. He may call 
one group good, another fair, and a very few, poor. This 
crudity is often due to the fact that it is extremely difficult 
to make an accurate classification mentally. In order to 
overcome this difficulty, the name of every person tested 
Was written on a small card. The pack of cards was then 
handed to the foreman and he was asked to divide all 
the men into three equal groups, according to their ability. 
This was comparatively easy. Having done this, he was 
f“ext asked to take each group and arrange the cards in 
in the same way. Thus, by dividing the work and by 
taking the mechanics of the classification simple and 
helpful, an unambiguous and probably more reliable 
result was obtained. The tests were then also ranked 
ar >d compared with the ranking of the men being trained. 
The correlations found were: plus .81 for the form-board 
test, plus .75 for the cube test, plus .84 for the Stenquist 
tes t, and plus .90 for the three tests combined. These 
correlations are unusually high, but they were based not 
0t i the foreman’s first impression of the men but his ma 
ture opinion after having been in close contact with them 
during several weeks of intensive training. (The foreman 
Ww nothing whatsoever of the performance of the men 
m the tests when he made his rankings.) The ranking of