ON THERMOPILES. 91 



the thermopiles. I am not aware that anyone has brought this 

 principle to bear either before or since. In the ordinary thermopiles 

 the number of elements is very great ; a plate of bismuth is soldered 

 to one of antimony, and you have as many as you please, and the 

 solderings all side by side form the surface on which the heat acts. If 

 you applied this to the moon the moon's light would shine on each of 

 the solderings and you would have some of the effects on all the 

 separate solderings. I found some difficulty in getting thermopiles of 

 sufficiently equal parts to use in the apparatus, and therefore I thought 

 of a simpler form which was within my own powers of manufacture. 

 I tried to make thermopiles with several alternations, but the matter 

 is one of great delicacy ; the bars are so thin and have to be handled 

 with such a care that a maker told me that after making them for 

 some years his sense of touch became so delicate, and he handled 

 everything so lightly, that he frequently, without intending it, let fall 

 heavier things. I adopted the plan of having one bar of bismuth and 

 one of antimony soldered to a copper disk. I made the bars finer 

 and finer every time, and the finer I made them the stronger was the 

 effect, because the cross section to carry away the heat and the mass to 

 be heated were smaller, and the limit of the thickness really was the 

 difficulty of manufacture. By having a thin disk of copper I was able 

 to work with much finer bars. I think this construction of a thermopile 

 is worthy the attention of instrument makers, because it so much simpler 

 to make. I do not see any reason a priori why as great an effect 

 should not be obtained from that construction as from fifty or sixty 

 pairs. The heat falls on the face of the copper which is a good con- 

 ductor, and if it is perfect I see no reason why there should not be nearly 

 as much power obtained from a single pair as from a great many. I do 

 not think there is anything else I need say on this subject, unless it is 

 to state the amount of heat obtained which I compared with that 

 obtained from a vessel filled with boiling water. The heat derived 

 from the moon would lead one to infer that the radiation was equal to 

 what it would be if the moon's surface at full moon was at about the 

 temperature of boiling water, or something like 200 Fahrenheit. Ot 

 course the assumption must be made that the moon has the same 

 radiant power as a lamp-blacked surface of a tin vessel of boiling 

 water, and that we have no right to assume at all. In fact I ought 



