310 THE POPULAR SCIENCE MONTHLY. 



abnormal conditions of pressure, of wind- direction, of cloud-formation 

 and movement." 



The most remarkable and interesting feature of the development 

 of tornadoes is the fact that they nearly always form southeast of a 

 moving center of low pressure, and their tracks, scattered here and 

 there, conform closely to the progressive direction of the main storm. 

 For example, on February 19, 1884, forty-four tornadoes occurred in 

 • Georgia, Alabama, and South Carolina, but principally in Georgia and 

 Alabama. They developed at a distance of from five hundred to two 

 thousand miles from a storm-center that moved across the northern 

 part of the United States, beginning at the northern extremity of the 

 Rocky Mountains in Montana, thence southeasterly through Dakota, 

 Minnesota, and Wisconsin to Northern Illinois and Indiana, northward 

 through Michigan across Lake Huron, disappearing north of Quebec. 

 This sudden sharp turn of the storm-center southAvard into Illinois and 

 Indiana seems to have relation to the uni^recedentedly large number of 

 tornadoes that developed not far from the South Atlantic coast, ex- 

 tending inland as far as Southern Illinois and Indiana. This south- 

 ward lunge of a mass of cold, moist air seems to have caused the ab- 

 normal conditions of temperature and dew-point, and the high winds 

 necessary to cause the most tremendous exhibition of destructive tor- 

 nado-power ever recorded by the Signal Service. This invariable 

 location southeast of the storm-center is one of the main peculiarities 

 of tornado development upon which the predictions depend. 



One of the best illustrations of the advance made in definiteness in 

 prediction during 1885 occurred on August 3d, in the instance of the 

 tornado at Camden, New Jersey, and at Philadelphia, Pennsylvania. 

 In October, 1885, the writer had sent a short communication to the 

 press of the country, advocating that tornado-signals of either safety 

 or danger be shown during certain seasons of the year at all telegraph 

 stations in States in which tornadoes are frequent. The gentlemen of 

 the press had generally favored the scheme, and one of the editors 

 wanted to knoAV if the tornado at Camden had been predicted. It 

 occuiTed at 3.20 p. m. (seventy-fifth meridian time), and was very de- 

 structive, involving a loss of about half a million dollars' worth of 

 property. The chart used by Lieutenant Finley shows that tornadoes 

 were predicted and their location marked upon the map for the States 

 of Delaware, Southeastern Pennsylvania, and for New Jersey. The 

 tornadoes actually occurred in these States about eight hours from the 

 time of the prediction, which was made on the basis of the 7 a. m. 

 (seventy-fifth meridian time) telegraphic reports. On that day heavy 

 wind-storms were predicted for Vermont, New Hampshire, Massachu- 

 setts, New York, Northern New Jersey, and Eastern Pennsylvania. 

 Nature carried out these predictions with as fair a degree of accuracy 

 and with as definite a conformity to location as could be expected at 

 the present primary stage of this science. 



