April 14, 1916] 



SCIENCE 



545 



6 mm. long, 3 mm. Tvide, and 3/1,000 mm. tMck, 

 placed centrally and level with the surface of a 

 circular nickel-plated copper plate 12 mm. thick, 

 75 mm. in diameter. The manganin strip is elec- 

 trically insulated from the copper plate by means 

 of thin strips of mica which come exactly to the 

 common surface of the plate and strip. Under- 

 neath the manganin strip are cemented two 

 thermo-elements of tellurium and platinum joined 

 in series, and whose cool junctions are embedded 

 in opposite halves of the copper plate. A polished 

 nickeled hemispherical shutter encloses the outside 

 of the glass hemisphere, and when it is open the 

 radiation from the sky passes through the hemi- 

 sphere, falls upon and is absorbed by the uppA 

 surface of the manganin strip. Thus the thermo- 

 elements are warmed and deflection of the gal- 

 vanometer connected with the apparatus would 

 ensue. But this is reduced to zero by means of an 

 auxiliary current supplied by a potentiometer ar- 

 rangement. (Having secured the balance by 

 means of the potentiometer circuit, the shutter is 

 now closed and a heating current is applied to the 

 manganin strip until the temperature is again 

 raised so that with the same potentiometer current 

 the galvanometer again stands at zero. In these 

 circumstances, as in the Angstrom pyrheliometer, 

 the energy of the current expended in heating the 

 strip is equal approximately to the energy of the 

 sky radiation which heated the strip before. This 

 apparatus has been used with excellent results on 

 the snow, the sky, the sky and the sun, and the 

 sun alone. In the latter case the instrument was 

 compared with a standardized silver-disk pyrheli- 

 ometer. Corrections having been made for the 

 inclination of the rays to the surface of the hori- 

 zontal sky radiation instrument, reflection of 

 glass, and imperfect absorption of the lamp-black 

 close agreement was found between the results de- 

 rived from the two kinds of apparatus. We are of 

 the opinion that with this apparatus the sky 

 radiation can be measured to within perhaps 2 

 per cent. The reflecting power of snow for total 

 solar radiation was found to be 70 per cent. In 

 using the apparatus for the measurement of noc- 

 turnal radiation the glass hemisphere is removed. 

 Upon the opening of the shutter the strip cools 

 and thereby a deflection is produced in the at- 

 tached galvanometer. This deflection, however, is 

 brought to zero by introducing in the strip a heat- 

 ing current such that the temperature is restored 

 to what it was before the shutter w'as removed. 

 It is plain that the instrument may also be used 



for the measurement of the radiation of inclosures 

 at fixed known temperatures which might be re- 

 garded as perfect radiators. We hope to make 

 experiments of this kind in the effort to aid in 

 the determination of the constant of Stefan's 

 fourth power formula for the radiation of black 

 bodies. 



Invisible Companions of Binary Stars: G. C. CoM- 



STOCK. 



A large proportion of the visible stars are 

 shown spectrosoopically to be accompanied by 

 companions not separately visible. In a very 

 limited number of cases, such companions have 

 been otherwise found. The presence of such in- 

 visible companions is possibly, or even probably, 

 a normal stellar attribute. Aside from spectro- 

 scopic investigation, and in a field not accessible 

 to it, the most promising method of search for 

 such bodies is to be found in the disturbances 

 produced by them in the motions of binary sys- 

 tems. This has been realized in a very few cases, 

 e. g., Zeta Cancri. The present paper suggests a 

 simple method of testing suspected cases of this 

 kind and shows by its application to Zeta Heroulis 

 that this star is probably a triple system in which 

 the relative masses are of the order 100: 10: 1. 

 The two smaller bodies are separated by only a 

 twentieth of a second of arc. 

 Theory of Electric Conduction in Metals: Edwin 



H. Hall. 



In July, 1914, the author publishedi a paper in 

 which he reached the conclusion that the so-eallea 

 free electrons have little to do with electric con- 

 duction in metals but have an important function 

 in thermo-electric action. In 1915 he made the 

 suggestions that the metal ions, — which are prob- 

 ably equal in numbers to the free electrons in a 

 metal — may be of great effect in electric conduc- 

 tion. The idea is that during a collision between 

 an atom and an ion an electron may be trans- 

 ferred from the atom to the ion by the action of 

 a potential gradient due to an externally applied 

 E.M.F., whereas in the collision of two atoms the 

 electron would not pass. It can be shown that a 

 comparatively small number of ions might serve to 

 maintain a very powerful current. Some progress 

 has been made in adapting this general theory to 

 the requirements of Ohm's law and the known 

 temperature relations of electric conduction in 

 metals. 



1 Proceedings of the American Academy of Arts 

 and Sciences. 



- In II Nuovo Cimento, the first number for 1915. 



