Metals at High Temperatures. 375 



lengths of the various bars before and after heating conse- 

 quently correspond to permanent variations in the lengths of 

 the bars. This fact was also several times confirmed by meas- 

 urements before and after heating, with a comparator. 



An accidental disturbance of the adjustment is not wholly 

 impossible however, and we therefore referred the expansion 

 to the length of the cold bar as observed immediately before 

 the heating rather than after, as the latter measurement could 

 not be made until the following day after the oven had cooled 

 down. The amounts of such slight differences as were ob- 

 served are recorded in their proper places. The preliminary 

 heating of the bar to remove mechanical strains and inhomo- 

 geneities in the metal is not taken into account of course 

 in this connection. The permanent change in the length of 

 a new bar after the first heating is often very considerable, 

 hence each bar was always first subjected to a single prelimi- 

 nary heating to the highest temperature to be used in the sub- 

 sequent measurements. 



It may be further mentioned that the length represented by 

 one turn of the micrometer screw was determined directly from 

 the readings at the temperature of the room, the interval be- 

 tween the marks on the bar having been determined independ- 

 ently in advance upon the comparator. 



The remaining details of the length measurement will appear 

 from the example given in detail further on. 



Measurement of the Temperature. — During the expansion 

 observations the temperature of the middle of the bar alone 

 was measured, a thermo-element whose bare hot junction lay 

 upon the bar at its middle point being used for the purpose. 

 The fall in the temperature toward the ends was determined 

 separately by observations at points on each side of the center, 

 the element being moved along the bar and the differences as 

 compared with the center measured by direct comparison. 



Two different oven tubes were used in the course of the 

 series of observations here described. The first was wound 

 with nickel wire 2 mm in diameter and proved fairly satisfactory, 

 but in the hope of increasing the uniformity of temperature 

 distribution a second tube was prepared with l-2 mm wire in 

 trifilar winding, which proved much better. We also tried 

 rewinding the first coil with the heavier wire, but soon became 

 convinced that although it is much more durable it is less well 

 adapted to heating such a long and slender tube. 



It is a matter of the greatest care to arrange the windings on 

 such a tube so as to produce uniform temperature distribution. 

 Some very painstaking trials yielded considerably poorer results 

 than those here communicated. On the second tube men- 

 tioned, and the most successful one we obtained, the windings 



