106 On a new Theory of the Aurora, 



potential, the force of repulsion would be greater than the 

 gravitation attraction to the earth, and it would fly off through 

 space. 



For these reasons I rejected the theory, and now believe 

 that the magnetism of the earth still remains, as before, one 

 of the great mysteries of the universe, toward the solution of 

 which we have not yet made the most distant approach. 



In connexion with the theory of the earth's magnetism, I 

 had also framed a theory of the Aurora which may still hold. 

 It is that the earth is electrified, and naturally that the elec- 

 tricity resides for the most part on the exterior of the atmo- 

 sphere — and that the air-currents thus carry the electricity 

 toward the poles, where the air descending leaves it — and 

 that the condensation so produced is finally relieved by dis- 

 charge. 



The total effect would thus be to cause a difference 

 of potential between the earth and the upper regions of the 

 air both at the poles and the equator. At the poles the dis- 

 charge of the aurora takes place in the dry atmosphere. At 

 the equator the electrostatic attraction of the earth for the 

 upper atmospheric layers causes the atmosphere to be in un- 

 stable equilibrium. At some spot of least resistance the upper 

 atmosphere rushes toward the earth, moisture is condensed, 

 and a conductor thus formed on which electricity can collect; 

 and so the whole forms a conducting system by which the 

 electric potential of the upper air and the earth become more 

 nearly equal. This is the phenomenon known as the thunder- 

 storm. 



Hence, were the earth electrified, the electricity would be 

 carried to the higher latitudes by convection, would there dis- 

 charge to the earth as an aurora, and passing back to the 

 equator would get to the upper regions as a lightning dis- 

 charge, once more to go on its unending cycle. I leave the 

 details of this theory to the future. 



Baltimore, May 30, 1879. 



Appendix. — Since writing the above, Professors Ayrton and 

 Perry's paper has appeared in full ; and I am thus able to point 

 out their error more exactly. Their formula at the foot of 

 page 406 is almost the same as mine ; but on page 407, in the 

 fourth equation, the exponent of n should be + \ instead of 

 — ^, which increases their result by about 600,000,000, and 

 makes it practically the same as my own. 



1'otterdam, July 13th. 



