18 REPORTS ON THE STATE OF SCIENCE.—1918. 
In order that impact tests on notched bars should be strictly com- 
parable, it is necessary to consider the form and dimensions of the test- 
pieces, the apparatus used in testing them, and the method of record 
and measurement. 
In regard to the form and dimensions of the test-pieces :— 
In tension tests it is clear that the sizes of the test-pieces cannot be 
constant, but the proportions have been established on the law of simi- 
larity which ensures that the test-pieces of different dimensions should 
be geometrically similar in order that the results obtained may be strictly 
comparable. 
Experience with impact on notched bars shows that the resilience 
referred to the area of the cross-section through the notch, when made 
on pieces 10 by 10 mm. and pieces 30 by 30 mm., are not strictly com- 
parable, as the results obtained by the large piece are uniformly greater 
than those obtained with the smaller test-piece, and the difference is 
greater the greater the resilience of the materials. The reason for this 
difference appears to be due to the fact that in the rupture of a notched 
bar we have two distinct stages in the process: (1) the general deforma- 
tion of the bar, which is proportional to the cube of the homologous dimen- 
sions ; and (2) the rupture of the piece through the notch, which is pro- 
portional to the square of the said dimensions. 
The work of rupture should then be expressed by two terms—the 
work of deformation proportional to the cube, and the work of rupture 
without deformation proportional to the square or the area of the cross- 
section through the notch. Professor Schiile has recommended that 
the faces of the test-pieces on opposite sides at right angles to the 
length of the notch should be polished, and the volume of the piece 
strained in impact, shown by the dull surface, should be used in express- 
ing the resilience. 
The tests made by the authors show that in the brittle shell steel 
there was very little deformation, and the resilience is well expressed 
by the area of the cross-section through the notch, whereas in the case 
of the alloy steels the deformation was considerable, and - was not 
confined to the area of the cross-section, but extended to some extent 
on each side of the notch. The method expresses with less accuracy 
the resilience of the more resilient material than that of the brittle or 
less resilient material when expressed in terms of the area through the 
notch, and relatively under-estimates the more resilient material. It 
should be noted that similarity in the form of the test-pieces should be also 
provided in the radius of the notch, and that of the supporting knife-edges. 
Test-pieces.—For notched bar test-pieces, two standard forms have 
been recommended by the International Association for Testing Materials 
(see figs. 1 to 6). 
A small test-piece, 60 mm. long by 10 by 10 mm. in cross-section, 
having a notch 2} mm. deep, rounded at the bottom to a radius of 1 mm.! 
This is easily prepared with a twist drill or a properly shaped milling 
cutter. A larger test-piece, 160 mm. long by 30 by 30 mm. in cross- 
section, having a circular notch 15 mm. deep, formed with a twist drill 
’ The dimensions given are not exactly those recommended, but are those actually 
employed in the tests made by the Committee. The recommended depth of notch 
is 5 mm, and the bottom radius 2 mm. 
