W. A. Norton— Mcperiments on Coniaci Resistance. 44S 



suitable length for clamping were used in the experiments. The 

 lower piece was clamped to a horizontal iron bar, which was 

 firmlj clamped to the vertical pillars of the testing machine 

 used in my former experiments on deflection and set, and was 

 also firmlj propped directly beneath the point where the con- 

 tact occurred. The other piece, | inch in length, was keyed 

 to the under surface of the lever used in the same experiments, 

 at the farther end. The weights were placed on a scale pan 

 resting above this on the lever, and vertically over the surfaces 

 in contact. The depressions of this end of the lever were deter- 

 mined by means of a micrometer screw, which gave the equal 

 elevations of the other end to within 45^^;^ of an inch. The 

 firmness of the lower contact piece and its support was fre- 

 quently tested by causing the weights to press directly upon it, 

 without the intervention of the lever. The small thermal error 

 of the apparatus was carefully determined and allowed for 

 -' • any perceptible change of temperature occurred dur- 



• glei • ' ..,.., 



mg any single series of experiments ; but the 

 taken to secure a nearly uniform temperature during the pro- 

 gress of the experiments. The weights employed, in the more 

 precise determinations, ranged from 2 ounces to 24 ounces. 

 The apparent surface of contact varied from gV of a square 

 inch to a mere point. The touching surfaces were in some in- 

 stances smooth, in others rough ; and in the contact of plate 

 glass with plate glass, highly polished. The decrement of con- 

 tact distance was noted whenever a weight was put on, and the 

 increment when the weight was removed, and in general the 

 average of the two taken. By this means the thermal error, 

 when the rise or fall of temperature was uniform, would be elim- 

 inated ; as well as any error that might result from a change in 

 the coefficient of the contact resistance, induced by the pressure 

 and not passing off when the weight was removed. That errors 

 from irregular variations of temperature, irregular variations of 

 the coefficient of molecular resistance, and accidental causes, 

 might be in a great degree eliminated, the mean of a consid- 

 erable number of separate determinations was obtained in each 

 case. A comparison of these means for sets of experiments dif- 

 fering in number, showed that the irregular and accidental 

 errors were generally small. The initial pressure was the same 

 in the different sets of experiments, and was very slight — be- 

 ing barely sufficient to secure a decided contact 



When a weight was applied the resulting diminution of the 

 contact distance was generally greater than the increase that 

 resulted from the removal of the weight The reverse very 

 rarely occurred ; though the increment was sometimes equal to 

 the decrement It therefore generally happened that there was 

 a slight contact set when the pressure was withdrawn. These 



