94 Scientific Intelligence. 



began to be appreciable at 265^, which wave-length is ade- 

 quately shorter than the observed end of the solar spectrum, 

 namely 290^. The instrument was set for the mercury line 292 

 •58^ and calibrated with the mercury arc spectrum. All measure- 

 ments were started at the sharp absorption line t, 299*45'"'". 

 With a slit width of 0*or nra the exposures took from a few 

 seconds up to 10 minutes. Nothing was gained by lengthening 

 the exposure time, because of the ever-increasing density of fog 

 due to light of greater wave-length scattered by the optical sur- 

 faces. The failure of Wigand's first experiments was caused by 

 this scattered light, and the success of his later attempts to photo- 

 graph shorter wave-lengths than had been previously recorded 

 depended on the recognition of this fact and the removal of the 

 disturbing cause. It was found that bromine vapor afforded the 

 required light-filter. The saturated vapor was confined in a glass 

 tube 20 cms long and 3*5 cms in diameter. The incidence end of this 

 tube was closed by a plano-convex quartz lens of 22 cms focal 

 length. The end of the tube nearer to the slit of the spectro- 

 graph was made of a plane-parallel quartz plate. The bromine 

 vapor absorbed completely all radiations from the yellow-green 

 as far as some distance in the ultra-violet. On the other hand, 

 it was entirely transparent in the interval 303^ to 265^. This 

 absorbing screen practically removed all undesired light and 

 greatly increased the contrast, sharpness and detail of the spec- 

 trograms. 



The best negatives taken at Halle (100 meters above sea level) 

 gave 289*73'"'" as the very shortest solar radiation recorded. On 

 September 9, 12 good negatives were obtained by the investi- 

 gator when in a free balloon at an elevation of about 9000 meters 

 (5*5 miles). All of these plates gave inferior limits of the solar 

 spectrum less than 290*75^'", the shortest wave-length recorded 

 being 289*60^. Thus the solar spectrum has been extended by 

 l*5« a ' a as compared with earlier data obtained by the photographic 

 method. The fact that the extreme ultra-violet limit of the solar 

 spectrum is sensibly the same at elevations of 100 and 9000 meters 

 shows that the conclusion of Cornu and others — that the limit of 

 the spectrum increases steadily towards shorter wave-lengths as 

 the altitude increases — is untenable. Also the lack of agreement 

 between the data obtained by other investigators (save Miethe 

 and Lehmann) was due, in part at least, to light scattered within 

 the spectrographs employed. A table is given of 52 absorption 

 lines from t 299*45^'" to 291*1 8^. The two most refrangible 

 lines are new, and they were recorded on the negatives taken at 

 Halle as well as from the balloon. 



Hartley has ascribed the abrupt ending of the transmitted 

 solar spectrum to the absorption exerted by ozone. If the limit 

 is due to any constituent of the earth's atmosphere then the 

 observed lack of dependence of this limit upon the altitude 

 requires the assumption that most of the absorber lies above 9000 

 meters. This hypothesis is not strengthened hy the fact that at 



