304 PROCEEDINGS OP THE AMERICAN ACADEMY. 



of hard wood. S, S, S, in the figure, are parts of this wooden support 

 which would have been cut through by a vertical longitudinal section 

 through the middle of the apparatus. Certain edges of the support which 

 do not form part of the section shown are indicated in the figure by 

 light dotted lines. Behind, to the left of, the block Bi B^^ is shown a 

 copper wire TFj, about 0.1 cm. in diameter, which extends through the 

 centre of a wooden rod r and bears against Bi B^, thus making a back- 

 stop for the pressure exerted at the other end of the apparatus. From 

 the block B^ B.2 another copper wire, W^, held in firm contact with Bo B,, 

 leads away. The wires Wi and W2 are parts of the thermo-electric cir- 

 cuit of the apparatus, and are in metallic connection with the terminals 

 of a galvanometer. B^ Bi is provided with a water jacket Ji J^, the 

 construction of which is indicated by certain lines in Figure 5 and in 

 Figures 6, 7, and 8. Thus, Figure 6 shows a vertical cross-section 

 through Ji Jx near the left end of ^1 B^, the dotted lines indicating cer- 

 tain edges not lying in this section. Figure 7 shows a vertical cross- 

 section through t/i ^1, through B^B-^. and through the thermometer T^ 

 (Fig. 5). Figure 8 shows a horizontal section through J^ J^ and Bi By. 

 The block B2 B2 is protected by a jacket quite similar to Ji J^. Wads of 

 cotton were used to protect certain parts of each block which were not 

 covered by the jackets. 



The course of the water through the apparatus is indicated by arrows. 

 Thus, at the left hand the streams enters at A^ passes down along the 

 bulb of Ti through B^ i?i, thence by a rubber tube, longer in fact than 

 in the figure, to J\J\^ thence by another rubber tube to j^-, and out at 

 E]^. The flow of the right hand stream is strictly analogous. Each 

 stream usually carried 20 or more grams of water per second. The 

 thermometers T-^ and To were the same that were used with the 15 cm. 

 bar of iron and in previous tests of thermo-electric junctions. Sets of 

 observations at a given mean temperature were made in pairs, one set 

 having T^ the warmer, the other set having To the warmer. 



It was necessary to give careful attention to the electrical resistance of 

 the column of short iron bars ; for it could not be safely assumed that 

 this resistance would be either small or constant. It was found, naturally, 

 to depend somewhat upon the magnitude of the pressure applied at the 

 ends of the column. In the experiments upon soft iron which we are 

 just now considering, the pressure was exerted by means of a compressed 

 piece of india-rubber tubing, not shown in Figure 5. In later experi- 

 ments, with cast iron, it was applied through a lever as in Figure 5, 

 the force F being exerted upon the end of the lever by means of a 



