126 Dr. Tyndall on Molecular Influences. 



brane becomes the proximate source of heat which is applied to 

 the left-hand face of the cube. The quantity transmitted from 

 this source, through the mass of the cube, to the opposite face, 

 in any given time, will, of course, depend on the conductivity of 

 the latter, and its amount may be estimated from the deflection 

 which it is able to produce upon the needle of a galvanometer 

 connected with the bismuth and antimony pair. G is a galva- 

 nometer used for this purpose ; from it proceed wires to the mer- 

 cury cups M M', which, as before remarked, are connected by 

 platinum wires with A and B. The galvanometer is a carefully 

 constructed and delicate instrument from the workshop of that 

 skilful mechanic, Kleiner, in Berlin. 



The sole use of the mercury in the chambers m and m! is to 

 secure good and equable contact ; when the chambers are filled 

 with pure mercury, and this is allowed to remain in them 

 throughout an entire series of experiments, it is certain that the 

 conditions of contact are perfectly constant, and thus the most 

 fruitful source of doubt and error is effectually excluded. In 

 rough experiments the chambers might be dispensed with, and 

 the bent wire itself might be brought into contact with one face 

 of the cube, while the other face might immediately press against 

 the bismuth and antimony. The result however of many hun- 

 dreds of experiments made with the instrument in this state, has 

 been to prove the impossibility of preserving the conditions of 

 experiment constant, and to compel me to devise some means of 

 avoiding the irregularities which exhibited themselves. The in- 

 strument just described meets the requirements of the case ; care 

 is necessary in the use of it, but when care is taken, an accuracy 

 is attainable by it which, I believe, has been hitherto unequalled. 



The action of mercury upon bismuth, as a solvent, is well 

 known; an amalgam is speedily formed where the two metals 

 come into contact. To preserve our thermo-electric couple from 

 this action, their ends are protected by a sheathing of the same 

 membrane as that used in front of the chambers m m'. 



Previous to the cube being placed between the two membranes, 

 the latter, by virtue of the fluid masses behind them, bulge out 

 a little, thus forming a pair of soft and slightly convex cushions. 

 When the cube is placed upon its supports and the slider is 

 brought up against it, both cushions are pressed flat, and thus 

 perfect contact is secured. The surface of the cube is larger 

 than the surface of the membrane in contact with it*; and thus 

 the former is always firmly caught between the opposed rigid 

 projections, the slider being held fast in this position by means 

 of the spring r, which is then attached to the pin p. The exact 

 manner of experiment is as follows :-— Having first seen that the 



* The edge of each cube measured 03 of an inch. 



