136 gray: temperature uniformity 



and tucked under alternate lengths of the ribbon. Then follows 

 another layer of micabeston and, finally, a wrapping of asbestos 

 cloth. As each layer is applied it is drawn down firmly by a 

 tight helical bandage of cotton tape, which is removed as soon as 

 the layer has been cemented in place by warming and then cool- 

 ing the tube. The final wrapping of asbestos cloth is secured 

 by a bandage of nichrome ribbon. The lengths of ribbon bound- 

 ing each edge of the mat and the two on either side of the middle 

 are left long enough at one end to serve as electrical terminals. 

 When the heater is wound, these terminals come out at opposite 

 ends of a diameter at one extremity of the tube, and they con- 

 nect the two halves of the winding in parallel. 



This method of winding possesses the following advantages: 

 (1) Any irregularities in the resistance of the heating ribbon are 

 distributed in such a way as to produce a minimum effect upon the 

 longitudinal distribution of temperature. (2) The winding is non- 

 inductive. (3) The maximum difference in electrical potential 

 that can occur between adjacent portions of the winding is lim- 

 ited to the drop in a length of ribbon twdce as long as the furnace. 

 This makes the danger of arcing by accidental contact consider- 

 ably less than in the ordinary form of inductive winding. (4) 

 If adjacent windings should accidentally touch, only a short 

 length of ribbon could be short-circuited. (5) Spaces for the side 

 openings down which the length-defining wires are suspended 

 can be provided without affecting the heat supplied per unit 

 length of the furnace tubes. (6) The location of the terminals 

 combines both convenience and safety. 



A furnace of the type here described requires considerable 

 time to attain thermal equilibrium. The lag, however, causes 

 no great inconvenience, because in any precise expansion deter- 

 minations sufficient time should be allowed for the specimen to 

 reach mechanical equilibrium, that is to say, for all unbalanced 

 internal stresses to disappear. The gain in temperature uni- 

 formity and constancy more than counterbalances any time lost 

 from avoidable lag. Nevertheless, it seems likely that a furnace 

 amply sufficient for most purposes could be made with consid- 

 erably less lag. In this connection it might be mentioned that 



