458 Serviss — Internal Temperature Gradient of Metals. 



tinned, this water was released, the drip became uniform, and 

 the temperature gradient steady. Only slight stirring between 

 the observations was necessary to keep the conditions sensibly 

 constant. A mechanical device, consisting of two endless 

 chains driven in opposite directions vertically through the ice, 

 has been installed to accomplish the stirring uniformly. 



Connections. — All of the wiring outside of the thermostat 

 and the copper tube was from the same bobbin of No. 18 cop- 

 per annunciator wire, with double cotton insulation. The 

 connections throughout were hard soldered, in the same man- 

 ner as the thermal junctions, except at the galvanometer coils 

 and at the points where the shift was made from direct circuit 

 through the galvanometer to a potentiometer. At these places, 

 copper binding screws were used to clamp the ends of the 

 wires in contact. The connections in the galvanometer were 

 undisturbed ; those at the observing table were cleaned daily 

 or oftener with infusorial earth, in addition to the care which 

 was always taken to avoid contact of the bared ends with con- 

 taminating objects, especially the fingers. Warming oue or 

 another of these binding screws produced no perceptible deflec- 

 tion in the galvanometer. The constant temperature room of 

 the laboratory was unfortunately occupied by another research, 

 but every precaution was taken to protect the apparatus from 

 changes of temperature. The doors and the windows were 

 kept closed, the galvanometer and the observing table were 

 further protected from air currents by screens, the copper 

 lead wires were wound thickly with strips of cotton batting, 

 and carried from one instrument to another over a wooden 

 frame. The resistance of the lead wires was 0*48 ohms at 20°. 



Commutators. — When the specimens were first placed in 

 position in the thermostat, the current from the thermopile 

 was sent directly through the galvanometer, shunted by a 

 standard coil, or with a resistance in series. The lead 

 wires were branched so that the current through the galvano- 

 meter could be reversed either by a four-point mercury com- 

 mutator or by a double throw copper switch. Both were 

 tried and the mercury commutator was found to be less dis- 

 turbed by local thermal effects ; at least they were more uni- 

 form and so more perfectly allowed for. The copper switch 

 was, therefore, removed from the circuit, so as to have no 

 branch wires projecting, and the mercury commutator used 

 exclusively. 



The mercury commutator was made from a new block of 

 mahogany, the mercury was freshly distilled, and the lead 

 wires amalgamated by cleaning thoroughly with emery cloth 

 and rubbing with the mercury ; no acid was used. This com- 

 mutator was placed in a small wooden box, and well protected 

 by two thicknesses of heavy asbestos paper and loose asbestos 



