August i, 1890.] 



SCIENCE. 



61 



I have, however, is a fair representation of the original. I 

 have enlarged Figs. 5, 6, and 7 from the Eichard barograph 

 to make them directly comparable with the barograph sheet 

 at Washington, given in Fig. 8. The barograph at Owens- 

 boro was a mile and a half from the nearest point of the 

 tornado. It has been supposed all along that a tornado 

 could not produce any effect on pressure more than a few 

 hundred feet from its centre, but here seems to be good evi- 

 dence of an effect over 7,000 feet away. The first sudden 

 drop of the curve with an immediate return is a little sin- 

 gular, and appears like the sudden drop found by others, 

 and explained by the effect of the wind. Whatever may 

 have caused this drop, there is no doubt of the rise after it ; 



rig. 5, tornado St. Louis, Mo. (Jan. 12, 1890, 5.16 p.m.); Pig. 6, tornado Owens- 

 boro, Ky. (March 27, 1890, 6.17 p.m.); Fig. 7, tornado Cincinnati, O. (March 

 27, 1890, 8.20 p.m.); Fig. 8, thunder-storm Washington, D.C. (June 22, 1890, 

 9.8 P.M.). 



and this rise is far more striking at Cincinnati, where the 

 storm crossed 59 minutes after leaving Louisville. Fig. 5 is 

 also given from the barograph at St. Louis during the tor- 

 nado of Jan. 12 of this year. This has already been referred 

 to. There is no doubt of a marked rise here. Fig. 8 is a 

 portion of the pressure-curve made at Washington, D.C, 

 during the prevalence of one of the most severe thunder- 

 storms that has visited the station. There was a steady rise 

 for an hour, and a less rapid fall after the centre had passed. 

 This storm came up and passed off very slowly, and was at- 

 tended by most brilliant lightning and a terrific wind. The 

 general similarity in these curves, excepting Fig. 6, is very 

 striking. We may hope to get more of such curves on both 

 sides, and possibly in the centre of tornado-tracks. 



Instructions for Observing Tornadoes. 



If any one thing has been emphasized in these pages, it 

 lias been the extreme need of more light on this whole ques- 



tion. The earlier investigators of the phenomenon were un- 

 trammelled, to a large extent, by preconceived opinions, and 

 it must strike every one that few substantial facts unknown 

 to them have been brought out since their time. It is to be 

 hoped that the number of those willing to aid in establishing 

 the facts and ferreting out the mysteries will be largely in- 

 creased, and it is for such these instructions are given. 



1. It is very essential that one divests himself of every 

 preconceived notion about the whirling, sucking, or any 

 other action of a tornado. There is the utmost danger of 

 seeing the tornado do what we think it ought to do. It 

 would be far safer, if one has an inclination to such views, 

 to deny that such a view is correct, and only to accept it 

 after most incontestable proof. 



2. Note the time of day carefully, and specify whether the 

 time used is Eastern, Central, Mountain, or some city time. 



3. The side of the track on which the observer stood should 

 he given. 



4. Note the appearance of the clouds in the distance, — 

 whether they roll upward, come together from the north 

 and south,— or any other phenomenon connected with them. 



5. Special note should be made of a cloud of dust and its 

 general appearance. 



6. See whether the funnel-cloud is visible at a distance, 

 or suddenly breaks into view on the approach of the tornado. 

 If possible, locate the first appearance of the funnel by a tree, 

 house, or object close by, and, after the tornado has passed, 

 measure the distance from the observer's position to that of 

 the tornado when first seen. 



7. If the observer is a thousand feet or more away from 

 the funnel, or cannot see distinctly trees or objects in it or 

 near it, he should not try to make detailed observations of 

 the whirl, or any thing else at the tornado. Fix the atten- 

 tion on the motion of detached clouds. Make every effort, 

 by comparison with trees and houses between the tornado 

 and the observer, to get its height, width, and speed: these 

 can be much better found out at a little distance than close 

 to the cloud. 



8. Make careful observations of all electric displays, the 

 appearance of balls of fire, the sound of thunder, the roar of 

 the tornado, etc. 



9. If the observer is within a hundred feet of the tornado 

 on the north side, he need have no fear, and may carefully 

 examine all objects fiying just above the ground. He should 

 note carefully the foot, the middle, and the top of the fun- 

 nel, to see whether any tree or object is carried to his right 

 as he faces the funnel. 



10. Note also whether any object is carried up straight in 

 the funnel, or whether it is borne along in the swift wind. 

 If there is an uprush in the tornado, it ought to be easily 

 told ; and the appearance will be very different from that if 

 the object or house is borne along by the wind, and after- 

 ward inclined upward. If there is a sort of explosive effect, 

 the upward motion will be more or less jerky, and not steady 

 as in a stream. On the whole, the evidence seems to show 

 something like an uprush, though it seems conclusive that 

 this is not due to a vacuum in the tornado, as many think. 

 It is very plain that nothing can be sucked into the tornado, 

 because of a partial vacuum there; but if it rushes in or up, 

 it must be borne or propelled by a stream of electricity, so to 

 speak, or by a rush of air. One of the best illustrations of 



