March i6, 191 6] 



NATURE 



57 



itube with an adjustable spark gap in series with it. 

 iXhe structure of the line was studied by means of 

 an echelon spectroscope, consisting of thirtv-three glass 

 plates, each plate being 948 mm. thick. The resolv- 

 mg power of the instrument at 4686 was 441,421, and 

 the distance between successive orders of the line was 

 0350 A.U. The line when excited by a condenser 

 discharge was found to be very broad and diffuse, and 

 two successive orders were only just separated on the 

 best photographs. In some experiments the spectrum 

 tube was immersed in liquid air, but no measurable 

 improvement in the sharpness of the line was obtained. 

 This result may also be expected on Bohr's theory, 

 as the atom is charged when emitting the lines. It 

 ;was therefore decided to excite the line by means of 

 ja direct current, keeping the drop of potential be- 

 tween the anode and kathode as low as possible. The 

 direct-current machine, which was connected through 

 a liquid resistance to the electrodes, could give a 

 voltage of 2000 and an output of one kilowatt. As 

 ifairly large currents were passed through the gas the 

 cylindrical spectrum tube was made large, and was 

 also provided with two heavy aluminium electrodes, 

 |one of which was concave and the other a hollow 

 tylinder. The tube was used in the end-on position. 

 j It was found that when the cylindrical electrode 

 {was made the kathode the light was almost com- 

 pletely confined to the space inside the cylinder, and 

 that it was very intense. Experiments were usually 

 Londucted at a pressure of i mm., and the voltage 

 between the anode and kathode varied in different 

 experiments between 280 and 400. Under these con- 

 ditions, although the ordinary helium lines were 

 strong, the 4686 line was comparatively faint, and 

 pxposures of about two hours were necessary for 

 pbtaining a satisfactory photograph, 

 i The line was found to be surprisingly sharp; in 

 fact, it was comparable in sharpness with the lines 

 of the ordinary helium spectrum, and much sharper 

 ihan the H/3 line of hydrogen, which was photo- 

 graphed at the same time'. In addition, all the photo- 

 Uraphs showed that the line was a close doublet, the 

 Tomponents having almost equal intensity. The best 

 photographs were measured up, and the distance 

 ^part of the two components was found to be 

 V094 ^•^- It is interesting to note that the separa- 

 tion to be expected on Rydberg's theory is 0-067 A.U. 



The structure of the hne is shown in the acconi- 

 banying photograph, which is explained by the dia- 

 gram. The dotted lines represent the different orders 

 pf the 4713 helium line and its faint component, and 

 the full lines the 4686 line. The doublet is repre-' 

 f^ented by an,, and a} is a higher order of a. 



I 



merfeld {Ray. Akad. d. Wiss., Munich, 1916) on the 

 structure of spectrum lines, which was based on a 

 remarkable generalisation of Bohr's theon.-. In this 

 paper he quotes certain unpublished results of Paschen 

 on the structure of several lines of the series 



■ These experiments were still in progress when there 

 ippearcd a very interesting theoretical paper by Som- 

 XO. 24.20, VOL. 97] 



'{<*'"#} 



the first member of which is the 4686 line, and also 

 of several lines of the series 



-■k''A:if 



-^.,1 



(?// 



which includes the Pickering lines, and also another 

 series of lines near the hydrogen lines, one member 

 of which, at 6560-4, was first observed by one of us 

 in a helium tube. 



In complete agreement with Sommerfeld's theory 

 Paschen found that the 4686 line consisted of three 

 components, each of which was accompanied by 

 fainter satellites, and the two stronger components 

 were separated by a distance one-fourth of that be- 

 tween the outer components. The values of the 

 separations in Angstrom units as obtained by Paschen 

 are not given in Sommerfeld's paper, but it is stated 

 that the ratios of the separations of the components of 

 the 4686 line to the separations of the components 

 of Ha agree with the values predicted by the theon.'. 

 Our result for the distance between the components 

 of the doublet also agrees approximatelv with the 

 value predicted by Sommerfeld for the separation of 

 the two strongest components. Since the appear- 

 ance of the paper we have re-examined all our 

 photographs to see if they show the presence of a 

 third faint component which we had missed. On our 

 best photograph we found near one of the higher 

 orders of the doublet, but not completely separated 

 from it, a faint line. If this line is a lower order of 

 the third component its separation measured from the 

 doublet is about 0-40 A.U., and it is situated on the 

 higher wave-length side, as is to be exf>ected according 

 to Sommerfeld's theory-. 



E. J. Evans. 

 C. Croxson. 



Manchester University, March 4. 



Ground Rainbows. 



Mr. a. E. He.ath asks (Nature, March 2. p. 5) 

 how gossamer which *' seems to be a kind of 

 spider web, comes to be spread over so large an 

 area." Mr. Heath need have gone no further 

 than Selborne to find the correct explanation, given 

 by Gilbert White 140 years ago : — " Nobody in these 

 days doubts that they (the cobweb-like appearances) 

 are the real production of small spiders- which swarm 

 in the fields in fine weather in autumn, and have a 

 power of shooting out webs from their tails so as to 

 render themselves buoyant." Possibly the first part 

 of the sentence was not true when Gilbert White wrote 

 it, seeing that it is not always the case to-dav. The thick 

 clouds of gossamer noticed by Mr. N. T. Porter when 

 out shooting in the early morning were noticed also 

 by Gilbert White in September, 1741, when "intent on 

 field diversions I rose before daybreak." If a more 

 recent account of gossamer is preferred it mav be 

 found in Fabre's "Life of a Spider." 



Charles J. P. Cave. 



Meteorological Office. South Farnborough. 

 March 7. 



