256 



gray: linear expansivity 



sive local heating. The left-hand half of the bar was not, however, so 

 uniformly heated as the right and was consistently warmer, the maxi- 

 mum rise of 2° occuring in the second 5 cm. from the left end. The most 

 plausible explanation is irregularity in either the heating coil or the 

 insulation at this place. It is to be noted that the thermal conduc- 

 tivity of such a large mass of iron as was used proved still insufficient 

 to produce a uniform distribution; and further that the distribution 

 observed at any one temperature did not differ essentially from that 

 observed at any other. Even making the outer ends of the plugs con- 

 siderably colder than the interior affected the temperature within but 

 little. 



3ooo ^ 



2ooo (J 



I ooo i 



loo" 



200° 



3oo° 



4-oo° 



5oo° 



6oo° 



Fig. 5. Expansion of annealed cold-drawn Bessemer steel 



Changes in this furnace are now in progress for securing a fuller appli- 

 cation of the principles discussed at the outset of this section. Two 

 concentric heaters separated by an insulating layer are provided, and 

 a recently devised scheme of longitudinal winding is introduced to 

 facilitate, among other things, uniform distribution of the heat supply. 



It is apparent that the end-heated plug, conducting trans- 

 versely while insulating longitudinally, affords a ready means 

 of preventing heat loss for any temperature of operation, and, 

 consequently, the entire interior, regardless of its length, will be 



