1006.] on Eclipse Prohhms and Observations. 248 



differently, under the polarisation tests, from other parts of coronal 

 streamers ? For our answer, let us look at the pictures. Here a 



streamer is shown, with edges nearly straight and parallel. Near its 

 base you will see certain curved arches, and a little further out the 

 arches are, as it were, burst through by the straight streamer, and 

 here are two remnants, each of which shows a point of inflexion. 

 That is an unpolarised picture. Now let us look at the polarised 

 pictures. These two pictures are polarised at right angles with one 

 another, and you will see the difference between the two. The two 

 inflected arch-remnants are equally strong in both, but the straight 

 streamer is markedly distinct in the one, and almost completely 

 obliterated in the other. The arches are not polarised, the straight 

 streamer is strongly polarised. The arches are probably incandescent 

 gas ; the straight streamer is possibly dust being driven outwards. 



I must hurry on to another set of observations made at Guelma, 

 with the assistance of Mr. Champion. This shows you the result of 

 photographing the spectrum of the light obtained from the corona. 

 There are two parallel bands of spectrum : the one corresponds with 

 the part of the corona verticaUy above the moon, and the other with 

 the part of the corona below the moon, whilst the dark interval 

 between them corresponds with the black moon. Such a photograph 

 would be got through an ordinary spectroscope. But here you see we 

 have a couple of such pairs of spectra, for the photograph was got 

 with a spectroscope in which there was a large double image prism in 

 train with the dispersing prism, the result being that one pair of 

 spectra is vertically over the other. 



The separated pairs of spectral images are polarised ; the one pair 

 is given by the Hght which is tangentially polarised, and the other by 

 the light which is radially polarised. Observe the difference in the 



R 2 



