578 BRIDGMAN. 



Stat with water at 100°. Instead, the maximum temperature was 

 usually set at about 99° and troublesome evaporation avoided by 

 covering the surface with a thin layer of oil. 



The manganin resistance gauge was calibrated from time to time 

 during the measurements, which extended over six months after the 

 preliminary work. There was a slow and uniform secular change in 

 the constant of the coil amounting in all to about f%. Of course 

 correction was made for this change. A new detail in the method of 

 calibration gave somewhat sharper results than possible by the method 

 described in the previous paper. By using petroleum ether to trans- 

 mit pressure to the freezing mercury instead of kerosene, it is possible 

 to avoid any effects due to viscosity of the transmitting medium, and 

 obtain results more quickly and sharply. The freezing pressure of 

 mercurv at 0.0° mav be reached from either above or below within the 

 limits of sensitiveness, about 1/3500, in fifteen minutes. The man- 

 ganin slide wire of the Carey Foster bridge was also repeatedly cali- 

 brated. This showed a secular change, due to wear, of about ^% 

 during the six months of the runs. 



The wires to be experimented on were usually wound either non- 

 inductively on themselves in the form of anchor rings of approxi- 

 mately 1.5 cm. external diameter, or were wound non-inductively in 

 a double thread cut on a bone core. It is essential that the method of 

 winding be such that the pressure is transmitted freely to all parts of 

 the coil without any mechanical hindrance from the frame on which 

 it is wound. This object is obviously at once attained when the wire 

 is wound on itself without a core, constrained only by a wrapping of 

 silk thread to keep it in shape, but this method is feasible only when 

 the wire can be covered with silk insulation without damage. If the 

 wire is soft like lead, it cannot be covered without damage, and it 

 must be wound bare on some sort of a core. Several attempts were 

 made before a suitable material for a core was found. At first, hard 

 rubber was used, but this is so compressible that at the highest pres- 

 sures the wire drops out of the grooves, and is so expansible that at the 

 highest temperatures the wire is stretched. A hard rubber shell on a 

 steel core does not work because the unequal compressibility of the 

 rubber and steel causes the rubber to crack. Bakelite was tried with- 

 out success. I was afraid to use mica because of the sharp bends 

 unavoidably introduced into the wire during handling. Finally bone 

 was found to be satisfactory from the points of view of both sufficiently 

 low compressibility and thermal expansion. A double 12 or 18 thread 

 of square section was cut in the lathe on the surface of a cylinder 



