238 



CARNEGIE INSTITUTION OF WASHINGTON. 



and also provides telephonic communication between the observer at the 

 telescope and the observer at the galvanometer. When an insulating layer 

 of cotton is placed around the thermo-couple cell the galvanometer is found 

 to be remarkably steady. With the registering device at a distance of 6.7 

 meters, a deflection of 0.14 mm. has been determined from 6 deflections, with 

 a probable error of =*= 0.010 mm., and similarly a deflection of 102.75 mm. 

 with a probable error of ±0.015 mm. It is therefore essential that the plates 

 be carefully measured on an adequate measuring-machine. 



It was thought at first that seeing conditions would not affect the results 

 to any marked extent, and in a measure this is true. The effect of poor seeing, 

 however, is to make the flat portions of the curves irregular instead of straight 

 lines, and this increases greatly the probable errors, especially for stars which 

 give large deflections. 



For the sake of simplicity the term "heat index" is introduced to denote 

 the difference between the visual and the radiometric magnitudes, the two 

 being assumed to coincide for spectral type AO. From the heat index, there- 

 fore, which is proportional to the ratio between the visual and the total radia- 

 tion expressed in magnitudes, we have a means of determining the relative 

 temperatures of stars, and hence a method for their spectral classification. 

 Similarly, the term "water-cell absorption" is used to denote the ratio ex- 

 pressed in magnitudes between the radiation transmitted by the water-cell and 

 the total radiation, and so indicates the distribution of energy in the infra-red 

 beyond 1.13 m- The 1-cm. water-cell used for this w^ork has been found to 

 transmit 90 per cent of the radiation incident upon it in the visual region, 

 but to cut off rather sharply at i.l3 ju. No radiation is transmitted beyond 

 1.34/z. The transmission curves of a considerable number of liquids are 

 now being investigated, with a view to finding one more suitable than water 

 for this purpose. The following table gives provisional values of the heat 

 index and water-cell absorption, uncorrected for atmospheric absorption, for 

 stars of various spectral classes, including Md variables: 



Some S- and R-type stars have been observed, but the deflections are ex- 

 ceedingly small, showing that they have little infra-red radiation compared 

 with stars of class Md. The remarkably large deflections given by stars of 

 type Md were first observed in o Ceti on December 6, 1921. The star was 

 then near minimum visual brightness, mag. 8.9, and gave a deflection about 

 equal to that obtained from a Virginis, a star of visual magnitude 1.2 but 



