204 Prof. H. L. Gallendar on Platinum Thermometry. 



Similarly in the case of the difference-formula (4) in terms 

 of pt, the maximum or minimum value of t is given in terms 

 of d' by the equation 



*(max.) = (l-dyiOOXp*/2= -(2500/d') (l-tf'/lOO) 2 . 



Dickson's Formula. — In a recent number of this Journal 

 (Phil. Mag., Dec. 1897) Mr. Dickson has proposed the formula 



(R+ a y= P (t+b) (6) 



He objects to the usual formula (3) on the grounds, (1) 

 that it leads to a maximum value of the resistance in the 

 case of platinum at a temperature of about £ = 3250° C, and 

 (2) that any given value of the resistance corresponds to two 

 temperatures. He asserts that " both of these statements 

 indicate physical conditions which we have no reason to sup- 

 pose exist/' In support of contention (1), he adduces a rough 

 observation of Holborn and Wien * to the effect that the 



* Wied. Ann. Oct. 1895 ; p. 386. Mr. Dickson and some other writers 

 appear to attach, too much weight to these observations of Messrs. Holborn 

 and Wien. So far as they go, they afford a very fair confirmation of the 

 fundamental principles of platinum- thermometry at high temperatures ; 

 but the experiments themselves were of an incidental character, and 

 were made with somewhat unsuitable apparatus. Only two samples of 

 wire were tested, and the resistances employed were too small for 

 accurate measurement. The wires were heated in a badly-conducting 

 inutile and were insulated by capillary tubes of porcelain or similar 

 material. The temperature of the wire under test was assumed to be the 

 mean of the temperatures indicated by two thermo-junctions at its 

 extremities ; but the authors state that " the distribution of temperature 

 in the furnace was very irregular." The resistance was measured by a 

 modification of the potentiometer method, and no attempt was made to 

 eliminate residual thermoelectric effects. Under these conditions the 

 observations showed that the resistance was not permanently changed by 

 exposure to a temperature of 1600° C, at least within the limits of 

 accuracy of the resistance measurements. It is quite easy, however, by 

 electric heating as in the " nieldonreter," to verify the difference-formula 

 at high temperatures, with less risk of strain or contamination or bad 

 insulation. (See Petavel, Phil. Trans. A (1898), p. 501.) 



The two series of observations (excluding the series in which the tube 

 of the muffle cracked, and the thermocouples and wire were so con- 

 taminated with silicon and furnace-gases as to render the observations 

 valueless) overlapped from 1050° to 1250° C, and showed differences 

 between the two wires varying from 10° to 45° at these temperatures, 

 the errors of individual observations in either series being about 10° to 

 15°. It must be remembered, however, that the two wires were of 

 different sizes and resistances ; they were heated in different furnaces ; 

 they were insulated with different materials; and their temperatures 

 were deduced from different thermocouples. Taking these facts into 

 consideration, it is remarkable that the observed agreement should be so 

 close. The observations at the highest temperatures in both cases, with 

 the furnaces full blast and under the most favourable conditions for 

 securing uniformity of temperature throughout the length of the wire, are 

 in very close agreement with the difference-formula (2), assuming d=l-75. 



The second specimen was also tested at lower temperatures, but the 



