230 MESSES. K. THRELFALL AND J. A. POLLOCK 



length is divided into two hundred parts, a movement of the slide over one part thus 

 corresponding to a change of temperature of one-hundredth of a degree. Each 

 division is 1 '73 millim. long. In order to avoid the difficulty which arises from the 

 alteration of resistance of plug contacts, we have made a slight innovation. We have 

 placed all the coils in series, and between each of them we have brought a stout 

 clamping terminal to the top of the box. Connection is made to the coils which it is 

 desired to use by means of stout terminals attached to flexible leads. The diagram 

 shows how the resistance of these leads is compensated. The "dummy" leads used 

 to compensate the thermometer leads are made too short by the length of the 

 flexible leads, which are, of course, made of the same wire as the dummy leads. 

 Assuming that the temperatures of the dummy leads and of the flexible leads are the 

 same, this gives a perfect compensation. 



The advantage of choosing the particular value of the box unit above mentioned 

 lies in the resulting fact that, setting aside corrections, the box is direct reading ; the 

 disadvantage is that with any other thermometers the box would lead to inconvenient 

 arithmetic. The apparatus is only intended to measure temperatures between 5 C. 

 and 35 C., consequently the large coil of the box has the same resistance as the 

 thermometer at 5 C. The contact on the bridge wire is also of manganin, to reduce 

 thermoelectric action, and there is no difficulty in subdividing the bridge divisions to 

 tenths by estimation, i.e., in reading to one-thousandth of a degree. We use one of 

 AYRTON and MATHER'S D'Arsonval galvanometers and find it very sensitive and 

 portable. 



The following constants will end our discussion of this matter : 



Legal ohms. 



Eesistance of platinum thermometer at 100 C 45'99173 



C. 34-32448 



Difference .\. .. . ... . H'66725 



Resistance of equal arms . .:. 10'027 



Testing Resistances. 



(1.) The resistance of each of the box units was ascertained in terms of the resist- 

 ance of one division of the bridge wire, the wire having been previously tested for 

 uniformity in the usual way. 



(2.) The same process was gone through with the " fives." 



(3.) The five box units in series were tested against each of the "fives" in 

 succession. 



(4.) The value of the large coil was ascertained in terms of the platinum ther- 

 mometer at C. through the intermediary of another box. 



We do not desire to dwell on this part of our work, as it is foreign to the main 

 object of this paper, but we must state that the comparisons were made with care and 



