346 PROCEEDINGS OF THE AMERICAN ACADEMY 



Weber, the differential current was .00 004 08 "Weber, which allowing 

 an increase of .004 in resistance for each added degree of tempera- 

 ture indicates* that the strip had been heated somewhat less than 

 0° 15 c. by the lamp radiation. A small (spherical-bulb) mercury 

 thermometer placed at the same point rose six times this amount. 

 Evidently only a small portion of the energy conveyed to the strip is 

 retained as increased temperature. The immensely greater part is lost 

 by re-radiation, conduction, and convection. This happens to the mer- 

 cury thermometer to a very much smaller extent, since the compara- 

 tively slow conveyance of heat between its outer and inner layers 

 enables it to retain a larger amount. 



The conduction from front to back of the thin strip is practically 

 instantaneous, and the equilibrium between heat received and heat 

 radiated is so soon established that the effect upon the galvanometer 

 is not increased perceptibly by prolonging the exposure after the 

 needle has reached the end of its swing. Hence the time of exposure 

 will, in general, be regulated by the sensitiveness of the galvanometer, 

 and will very rarely exceed eight to ten seconds. The strip itself takes 

 up and parts with (sensibly) all its heat in a fraction of one second. 



This promptness in the action of the metal strip gives it a great 

 advantage over the thermopile for measures of precision. But, beside 

 this, the deflection produced by the single strip and bridge is greater 

 than that from the thermopile, if the element of time enter into the 

 comparison, and still more if the relative areas exposed to radiation be 

 considered. 



Although (for the reasons just cited) far from as sensitive as we 

 can make it, such a strip then is yet more sensitive than the pile. A 

 number of thermopiles, selected as the most sensitive in the writer's 

 collection, have been exposed to the same source of radiation, placed 

 at the same distance as in the previous experiments. They were di- 

 rectly connected with the uushunted galvanometer and enclosed in 

 various cases, as follows : — 



A. Large thermopile, by Elliott (Tyndall-lecture pattern), composed 

 of sixty-three couples, on customary stand, but without cones. Face 

 blackened. Area of working face = 15 mm. by 16 mm. =240 sq. 

 mm. Internal resistance 5.90 ohms. 



B. Very sensitive thermopile of extra small elements ( 1 6 couples) 

 with cardboai'd diaphragm ajjerture G.6 mm. diameter. Area of work- 

 ing face (circle G.6 mm. diameter) = 34 sq. ram. Internal resistance 

 0.97 ohm. 



* See Forimila, page 355. 



