Dixon — A Thermo- Electric Method of Cryoscopy. 55 



support of the thermo-couple loads, connection was established between the 

 two pairs of leads, and tlie circuit was complete. By releasing the clip and 

 rotating it round a vertical axis through 180°, and clamping it again on 

 the support, the current from the couple may be reversed through the 

 galvanometer. 



In this form of reversing key the junctions being of the same metal and 

 — if desired — made of the same piece of metal, thermo-electric effects set up 

 by temperature-difference at the junctions are reduced to a minimum. 

 Notwithstanding this, it was found that these differences of temperature were 

 a source of error. To maintain the junctions on the opposite sides of the 

 support at the same temperature and so eliminate the error, these connections 

 were made underneath liquid petroleum, contained in a beaker, on the 

 bottom of which rested the support of the thermo-couple-leads. The 

 petroleum was kept stirred during observations. 



It will be found convenient to have the galvanometer leads a considerable 

 length, so as to allow a suitable distribution of the parts of the apparatus ; 

 consequently it is essential that they sliould have a sufficiently large cross- 

 section, so as to offer but a small resistance ; otherwise changes in tem- 

 perature, from which it is impossible to shield them, wijl alter the 

 sensitiveness of the apparatus. With the key described there is no 

 objection to having the galvanometer leads of diiferent copper wire and 

 heavier than those coming from the junctions. 



Some special precaution is also needed to secure that the junctions at 

 the binding-screws and tliose in the galvanometer are at the same tempera- 

 ture. In the case of these connections it is all the more necessary, because 

 elements of the junctions are of different materials — viz., brass and copper. 

 It was found that the different temperatures of the opposite sides of the 

 galvanometer in an ordinary laboratory could cause quite an appreciable 

 deflection. To remove this the galvanometer was placed in a thermostat, 

 arranged to maintain a temperature of about 21° C. For this purpose I 

 used one of Hearson's incubators witli a hole cut in the wooden door 

 through which the beam of light illuminated the galvanometer-mirror, and 

 was reflected back to the scale. The inner glass door was found not 

 to injure the sharpness of the image of the cross-wire sufficiently to be 

 objectionable. I found it necessary to stand the galvanometer on a stout 

 glass plate on the copper floor of the thermostat, which otherwise slowly 

 sagged under the pressure of its feet. The thermostat during my obser- 

 vations, extending over a year, was maintained at temperatures which varied 

 very slowly between 20-2° 0. and 21-5° C, so that at any moment the parts 

 of the galvanometer must have been very closely at the same temperature. 



