THE HK AT ING OF TRANSFORMERS 185 



increase of resistance, is less than that obtaiiu d by thermo- 

 metric measurements. This is because the averag. tempera- 

 ture of the winding* is less than that of their upper portion. 

 Thus thermometric measurements may sometimes afford a 

 nb. re correct indication of the maximum temperature in any 

 part of the winding than measurements based on the increase 

 of resist ance of the windings. However, the small depth of 

 oil in sinall-si/ed oil-immersed transformers is rarely sufficient 

 to occasion this result, and in small sizes the resistance method 

 will usually give a more correct indication of the maximum 

 temperature. An objection to the resistance method of 

 '.mining the temperature arises in the case of transformers 

 where the windings are divided into sections which are con- 

 nected in series. In such a case the temperature calculated by 

 resistance measurements is the mean temperature of all the coils, 

 and therefore does not afford reliable indication as regards the 

 temperatures of individual coils, some of which may be operating 

 at temperatures much higher than the mean temperature. 



Comparison <>/ r<iri<>nx tiff's </' trtBuformcr$ <U mianl* In-atinij. 

 While either the shell or the core type of transformer IUMV 

 be so designed as to conform to any reasonable specification 

 with respect to temperature rise, I nevertheless consider that 

 th. .xln-11 type is distinctly inferior to the core type as regards 

 adaptability to oil immersion. In either type, ducts through 

 which the oil shall circulate may be readily provided, but 

 there would appear to be no such natural arrant m< nt possible 

 for the sh.-ll type as that of which our 20-kva transformers of 

 . I!', 50, 51, 58, 54 and 55, are typical examples. P. th 

 and core are well Lathed with the circulating oil. In the 

 shell type the coils are less readily cooled than the core unless 

 valuable space be sacrificed in the winding window. The 

 liar-shell type represented diagrammatic-ally in I igs. 8 

 ami 1 is still more greatly inferior as regards the difliculty of 

 cooling the copper, and it is significant that it is the exploiters 

 of the circular-shell type who have displayed the greatest 



