EARTH AND SUN AS MAGNETS HALE. 153 



the rods projecting from the furnace are connected to a galvanometer. 

 Harker found that when one of the carbon terminals within the fur- 

 nace was cooler than the other a stream of negative electrons flowed 

 toward it from the hotter electrode. Even at atmospheric pressure 

 currents of several amperes were produced in this way.^ 



Our spectroscopic investigations, interpreted by laboratory experi- 

 ments, are in harmony with those of Fowler in proving that sun 

 spots are comparatively cool regions in the solar atmosphere. They 

 are hot enough, it is true, to volatilize such refractory elements as 

 titanium, but cool enough to permit the formation of certain com- 

 pounds not found elsewhere in the sun. Hence, from Harker's ex- 

 periment, we may expect a flow of negative electrons toward spots. 

 These, caught and whirled in the vortex, would easily account for 

 the observed magnetic fields. 



The conditions existing in sun spots are thus without any close 

 parallel among the natural phenomena of the earth. The sun-spot 

 vortex is not unlike a terrestial tornado, on a vast scale, but if the 

 whirl of ions in a tornado produces a magnetic field, it is too feeble 

 to be readily detected. Thus, while we have demonstrated the ex- 

 istence of solar magnetism, it is confined to limited areas. We must 

 look further if we would throw new light on the theory of the mag- 

 netic properties of rotating bodies. 



This leads us to the question with which we started: Is the sun 

 a magnet, like the earth ? The structure of the corona, as revealed at 

 total eclipses, points strongly in this direction. Remembering the 

 lines of force of our magnetized steel sphere, we can not fail to be 

 struck by their close resemblance to the polar streamers in these 

 beautiful photographs of the corona (fig. 11) taken by Lick Observa- 

 tory eclipse parties, for which I am indebted to Prof. Campbell. 

 Bigelow, in 1889, investigated this coronal structure and showed that 

 it is very similar to the lines of force of a spherical magnet. Stormer, 

 guided by his own researches on the aurora, has calculated the tra- 

 jectories of electrons moving out from the sun under the influence 

 of a general magnetic field and compared these trajectories with the 

 coronal streamers. The resemblance is apparently too close to be the 

 result of chance. Finally, Deslandres has investigated the forms and 

 motion of solar prominences, which he finds to behave as they would 

 in a magnetic field of intensity about one-millionth that of the earth. 

 We may thus infer the existence of a general solar magnetic field. 

 But since the sign of the charge of the outflowing electrons is not 

 certainly known, we can not determine the polarity of the sun in 

 this way. Furthermore, our present uncertainty as to the propor- 

 tion at different levels of positive and negative electrons and of the 



1 King has recently found that the current decreases very rapidly as the pressure in- 

 creases, but is still appreciable at a pressure of 20 atmospheres. 



