758 BELL SYSTEM TECHNICAL JOURNAL 



to coiling, this condition is often encountered. Furthermore, in 

 very thin materials there is a side flow so that the upper sheets fail 

 to take their share of the load and tend to curl up around the pene- 

 trator in such a manner as to prevent the proper measurement of 

 depth. There is also a possibility of including small particles of dirt 

 between the specimens which would result in lower hardness readings. 

 These matters must be very closely watched. An indentation in the 

 anvil just barely visible on a polished surface in strongly reflected 

 light will cause erroneous readings. It is necessary, therefore, that 

 the anvils be sufficiently hard for the purpose. During the past year, 

 harder anvils have been furnished and greatly improved results have 

 been obtained in the testing of thin materials. 



A comparison of the methods used by the laboratories in making 

 tests indicated that personal factors might account for some of the 

 differences in readings. One of these personal factors which appeared 

 to be of major importance was the amount of time allowed for the 

 drift of the penetrator after the major load is completely applied. 

 It was the custom of some of the participating laboratories to allow 

 this drift to continue until it had practically ceased while others 

 removed the major load and read the hardness as soon as possible 

 after the major load had been completely applied. This drift is very 

 noticeable in the softer tempers of non-ferrous metals and may amount 

 to as much as 10 Rockwell numbers. In general, it is more noticeable 

 with non-ferrous metals than with steels. In order to eliminate this 

 personal factor, the practice of operating the reset mechanism which 

 removes the major load, immediately upon the full application the 

 major load was adopted. 



Three separate rolling series were used in the round-robin tests 

 which were made after the various machines had been checked up. 

 In each case tests were made with both ^e and ^-in. balls using the 

 100-kg. load. All tests were made on each of six machines located in 

 the six different laboratories. Typical results of one of these series are 

 shown in Fig. 2 which gives the results of a brass rolling series using 

 both the \ and i^-in. ball. It will be noted that the greatest sensi- 

 tivity is obtained by using the i^-in. ball, and this was later adopted 

 as a standard ball for testing brass. Much better agreement between 

 Rockwell testers than between scleroscope testers is also apparent. 



Rockwell data were obtained on the high-brass and clock-brass 

 rolling series in all thicknesses and tempers using one thickness of 

 material and the standard "B" scale. The plot of the results of these 

 tests^on_^high_brass is shown in Fig. 6. These curves appear normal 

 for thicknesses of 0.012 in. and above, but below this thickness the 



