276 Rotary Action of the Providence Tornado. 



taining a violent whirlwind action, must come in horizontally, 

 and in the same gradually involuted courses ; or, must descend 

 in like manner from a higher region, in and around the outward 

 parts of the whirling cone. I have long since been led to believe 

 that this impulsive accession comes from both these sources, but 

 chiefly from the latter ; and that this motion of accession and 

 support is spirally downward in the outward portions of the 

 whirl. The latter being, in its higher portions, often greatly ex- 

 panded, as noticed by Mr. Allen. 



The evidence on which this opinion rests, can be but partially 

 alluded to here ; but I will suggest the following considerations : 

 — 1. The ascertained existence of a stratum of unusually cold 

 air in the higher region of clouds, on some particular days re- 

 markable for the occurrence of numerous thunder gusts and tor- 

 nadoes :* 2. The observed descent of a portion of the clouds in 

 front of the nucleus or body of a heavy squall or tornado, which 

 may sometimes be traced by the eye as low as the existing limit 

 of condensation will afford opportunity for observation : 3. The 

 fact noticed by Mr. Allen and others, that adjacent "to the exte- 

 rior edge of the circle of the tornado" or whirlwind, the previous 

 breeze often continues " to blow uninterruptedly from the same 

 quarter" as before :f 4. The last fact, when taken also in con- 

 nexion with certain peculiar and striking effects in the outward 

 portions or edge of the tornado, a knowledge of which I have 

 gathered from various sources : 5. The coldness of the air which 

 has been noticed at the edge of a whirlwind : 6. The instant 

 penetration of the lower end of the whirlwind into thick forests, 

 and into hollows and ravines, which has been frequently noticed : 

 7. The direct memorials of downward action in the outward por- 

 tions of the whirl which 1 have myself met with, on the tracks 

 of different tornadoes. 



* This change of upper temperature I think can be clearly made out on the day 

 of the New Brunswick tornado, which was but one of many tornadoes and thunder 

 gusts w^hich appeared in this part of the United Stales on the same day ; and on 

 the preceding day in Illinois and other western states. 



In the New Haven Gazette are accounts of five severe tornadoes which occurred 

 in the states of New Jersey, Connecticut, Massachusetts and Rhode Island, on the 

 afternoon of August 15, 1787. I can also refer to many more recent cases of this 

 kind. 



i The observation here quoted is one of many which show the error of the 

 very hasty generalization which alleges a circuit or annulus of calm air to have 

 been observed on all sides of tornadoes and hurricanes. 



