114 Trans. Acad. Sci. of St. Louis. 



nadoes and tropical cyclones may produce much more 

 wide-spread effects. A wind disturbance among the 

 atmospheric ions which sometimes accumulate along the 

 jnagnetic lines of force at the earth's magnetic poles, 

 might be expected to produce the effects which have 

 long been observed. 



On ten or twelve occasions similar effects were ob- 

 served, which were caused by a local dash of rain. When 

 rain falls continuously or at intervals during a day when 

 the sky is covered with clouds, which extend over ad- 

 joining states, the needle shows no appreciable disturb- 

 ance. The horizontal component of the earth's field is 

 then much weaker than it would be if the sky were clear, 

 but there is no additional change due to a rainfall. The 

 limiting condition has been already reached. 



When small clouds are scattered over the sky and a 

 local fall of rain occurs at the observing station, the 

 sunlight passing through the air through which the rain- 

 drops fall, a very marked magnetic storm is produced. 

 Such a disturbance is represented in Plate XXII. 



This rain, which was very violent, began while I was 

 at the noon-day meal. It lasted about ten minutes. 

 When the tent was reached the rain had practically 

 ceased at the station. The sun was visible during most 

 of the time while the rain was falling. The area cov- 

 ered by the rain was probably not over one or two square 

 miles. Its boundary could be seen upon the lake to the 

 south, while I was on the way to the tent. The needle 

 showed that we were then in the midst of one of the most 

 violent magnetic storms of the summer. The vibrations 

 ceased about five minutes after observation began. The 

 extreme reading of the scale for each oscillation was 

 taken. They are represented in Plate XXII, together 

 with subsequent readings of the needle represented on 

 the same time scale. Readings made before leaving the 

 tent for lunch are also shown. The gap between the 

 readings before and after lunch is only in part repre- 

 sented, as is indicated by the figures at the bottom of the 



