202 Holborn and Day — Gas Thermometer at 



possible even with inhomogeneous wires to carry out accurate 

 temperature measurements by always working under the same 

 conditions. It should be borne in mind however that this 

 accuracy is only relative and that the same thermo-elements 

 with a different temperature distribution between the junctions 

 may show considerable variations. 



To obtain the limits of absolute accuracy the fall in temper- 

 ature along the wire must be varied as much as possible. 



In the investigation here presented this was accomplished in 

 two wa} 7 s, either by measuring the temperature at a given 

 point with two elements having a common hot junction and so 

 arranged that the fall in temperature along the wire from the 

 hot to the cold junctions could be varied ; or secondly, by using 

 as many different points of an element as possible successively, 

 as junctions, and comparing them in a very short oven under 

 exactly similar conditions of temperature distribution. 



For example, b} 7 the first method the temperature near the 

 end of a long electrically heated tube would be measured with 

 two elements joined at their hot junctions, and the one passing 

 out at the end near by, while the other extended through the 

 middle of the tube which is hotter than near the ends, and out 

 at the far end. It is clear that the distribution of temperature 

 from hot to cold junction along the wires of the two elements 

 would be very different and could be varied at will by simply 

 moving the common junction to and fro. 



Any lack of homogeneity in the wires would then become 

 apparent at once in differences between the readings of the two 

 elements. 



We rather preferred the second method, the small melting- 

 point oven (fig. 3) being well adapted to the purpose. 



The wires of the elements are best simply bound together 

 with a short platinum wire when it is desired to use several 

 points provisionally as hot junctions. It has however no 

 influence upon the thermo-electric force if the wire be cut in 

 different places and afterward joined in the oxy-hydrogen 

 flame. 



Thick lumps along the wires are in the way to be sure, but 

 with a little practice a junction can be made which can hardly 

 be detected. 



Thermo-elements as they are now prepared in a diameter of 

 08 mra are very homogeneous and over a length of several 

 meters show no differences which could affect the temperature 

 measurement 1°. Thin wires reduced from the above by 

 drawing are often less uniform. Two wires, for example, 

 which had been drawn down to 0*25 mm showed about double 

 this variation, nor could this difference between the thin and 

 thick wires be removed by glowing electrically. 



