Second Series. — Thermo-Multiplier. 449 



to that of the pile.* The length of the conic frustum which I 

 employ is If inches, and its aperture If inches. 



7. It is well known that the deviations of galvanometers are 

 not, generally speaking, proportional to the forces producing them, 

 and that for the most part angular spaces at greater distances 

 from zero correspond to increments of force greater than for 

 equal spaces near zero. Thus to cause the needle to advance 

 from 25° to 30° requires a force greater than to make it deviate 

 from 0° to 5°. Also the force indicated by a deviation of 30°, is 

 more than six times the force indicated by a deviation of 5°. M. 

 Mellon i has pointed out an ingenious method of comparing the 

 values of the different parts of the scale. This consists in em- 

 ploying two constant sources of heat to affect the opposite ex- 

 tremities of the pile, and after observing their separate effects, 

 noting their joint effect, which will not generally be equal to the 

 arithmetical difference of the others. Thus let one source of 

 heat force the needle in a positive direction to 30° on the scale, 

 and a second source of heat acting separately produce a negative 

 deviation of 25°, the effect of both acting at once will not be a po- 

 sitive deviation of 5° merely, but probably will indicate some 

 greater number, as 6° or 7°. Thus, a true scale of degrees equal 

 in value to those near zero may be constructed. The execution 

 of this investigation is not so simple as it appears, chiefly owing 

 to a tendency of the zero point to shift during experiment, ap- 

 parently owing to a permanent electro-magnetic condition of the 

 conducting wire. It seems that this electro-magnetic condition 



* Suppose that we wish to have a conical reflector ABCD, such that the whole 

 of the parallel rays which fall upon it shall D 



reach some part of the surface AB of the 

 pile, which is all that we want, we have this 

 simple construction. Let the length of the ..--'' 



trumpet-mouth AE be given. Make FB f^ 

 equal and parallel to it. Join FA, and pro- 

 long the line to D, then is DAE the great- 

 est inclination that the sides of the cone can have to answer the purpose intended. 



