of Iron and German Silver. 487 



pipes k a scoop-like piece of sheet brass was soldered for the pur- 

 pose of diverting the entering water to the end-surface of the 

 bar. During the experiments the ends k were connected with 

 the pipes S, and the ends X with pipes which carried off the steam 

 or water, by means of caoutchouc tubing. 



It follows from the theory above given that the most favour- 

 able circumstances for the determination of K occur when the 

 ends M and N of the rod have simultaneously different tempe- 

 ratures — M being heated and N cooled during the first period, 

 M cooled and N heated during the following period, or vice versa. 



In order to regulate with facility the supply of the water and 

 steam according to this arrangement, the levers which turned 

 the plugs of the watercocks C and D, on the one hand, and 

 those of the steamcocks A and B, on the other, were connected 

 by a bar; so that by the motion of the two bars the four plugs 

 could be put into their right positions. All the cocks were pro- 

 vided with stops, and consequently could only turn within their 

 normal positions. With such an arrangement, it is easy to see 

 how the four cocks could be regulated at a distance. It was 

 only necessary to fasten cords to the ends of the connecting bars, 

 two of which were stretched by weights, whereby the plugs took 

 one of their normal positions ; and when the cords at the other 

 ends were pulled, the plugs were turned so as to take their 

 second position. In order to give the cocks a firm support, 

 each two of them (x\ and D, and C and B) were let into a piece 

 of wood, by which all radiation of heat to the bar was avoided. 



Instead of^four cocks, two four-way cocks might have sufficed, 

 as used by Angstrom in his researches. But then such a cock 

 would have been traversed at the same time by steam and by 

 water; and, on the one hand, steam would have been condensed, 

 while, on the other, the temperature of the water would have been 

 raised. By the above arrangement this inconvenience was 

 avoided. The temperature of the water was measured by a ther- 

 mometer inserted in the water-pipe. 



In order to observe the distribution of temperature in the bar, 

 in the first place a thin iron wire and a thin German-silver wire 

 were soldered in the middle of it, L, opposite to each other. The 

 ends of these two wires were soldered to two copper wires which 

 led to a galvanometer, and formed with the copper wires points 

 as shown in fig. 5. Between the points was the bulb of a sen- 

 sitive thermometer. The thermometer and wires were retained 

 in this position by a suitable keeper made of wood. If now the 

 middle of the bar and the points (which are in the surrounding 

 air) have different temperatures, a thermo-current ensues, the 

 intensity of which may be supposed proportional to the difference 

 between the temperature of the middle of the bar and that of the 



