i6 



SCIENCE. 



[Vol. XVI. No. 388 



and many also were straight blows with little violence. 

 Often a whirling cloud does not reach the earth, but it is 

 called a tornado. lu one list the Wallingford (Connecticut) 

 tornado of Aug. 9, 1878, is put down as only a heavy thun- 

 der-storm, although it ranks among the severest of all, for 

 there were thirty killed, fifty houses destroyed, and $250,000 

 of damage done. There will be given later a list of forty- 

 eight of the most severe tornadoes since 1873. No pains 

 have been spared in making this list as complete as possible ; 

 and while in a few cases the loss may be slightly overesti- 

 mated, yet in the main it is believed to be accurate. 



Hinrichs's Views. 

 Mr. Hinrichs of Iowa has devoted many years to the study 

 of tornadoes, and has given careful attention to the subject 

 of a classiQcation. A resume of his views has been given 

 in the American Meteorological Journal (vol. v. pp. 306, 

 341, and 385). He would make a careful distinction between 

 a tornado and a squall, or derecho, as he would call it. The 

 latter he describes as a straight-line heavy wind, usually ad- 

 vancing from the north-west for many miles across the State. 

 While it is frequently destructive, yet it is of an entirely 

 different type from the tornado. It is barely possible that 

 there has been a slight confusion here, and that the apparent 

 movement of the derecho from north-west to south-east is 

 complicated by the occurrence of a series of tornadoes or de- 

 structive thunder-storms running in parallel lines from south- 

 west to north-east. It hardly seems probable that we have to 

 deal with a particular class of storm (derecho) in Iowa that 

 is not found east of the Mississippi. What Mr. Hinrichs calls 

 the "front" of the derecho may be either a thunder-storm or a 

 tornado, and the universal law controlling these is a move- 

 ment from south-west to north-east. We have already seen 

 that these violent storms travel in parallel lines, beginning 

 earlier in the afternoon at a point toward the north-west of 

 a tornado region, and gradually working southeast. Each 

 line occurs later and later, and it is also a fact that fre- 

 quently the later storms have the longer paths. These ap- 

 pearances might also occur in a derecho. 



Hinrichs's Classification. 



The following shows the classification of tornadoes accord- 

 ing to Hinrichs's method : — 



A. Notable Tornadoes. 



Class I. Multiple. 



(a) Large. 



(6) Small. 

 Class II. Single. 



(a) Large. 



(6) Small. 



B. Minor and Doubtful Tornadoes. 



Finley's Classification. 



Lieut. Finley would' make no effort at a definite classi- 

 fication according to violence or extent, but would take 

 every funnel-shaped cloud, whether it reaches the earth or 

 is seen in the clouds, and give it a county, date, time of oc- 

 currence, direction of motion, shape of cloud, and width of 

 path. 



Objections. 



Neither of these methods seems satisfactory for our pur- 

 pose. The first is quite involved, and, moreover, unites two 



tornadoes under the same head, though they may be twenty 

 miles apart, provided they occurred on the same date. The 

 motion of a tornado is complex: it oftentimes lifts, and 

 passes a dozen miles before again striking the earth. Again, 

 we have several occurring in parallel paths on the same 

 date. To unite these together under the head of a "multiple 

 tornado" seems very injudicious. It will serve to great defl- 

 niteness if we consider each descent of a tornado-cloud, and 

 each occurrence of destruction in parallel bands, as a sepa- 

 rate tornado. My meaning should be distinctly understood. 

 I would call the tornado of St. Cloud and Sauk Rapids, 

 Minn., a single occurrence, because the cloud did not lift 

 between these two; also the Louisville and Jeffersonville 

 tornado was a single occurrence, for the same reason. I 

 would call the Fayetteville (Arkansas) and Marshfield (Mis- 

 souri) tornadoes separate occurrences, though having the 

 same date, April 18, 1880. They were a great many miles 

 apart, and, while undoubtedly in the same tornado region, 

 yet they should be kept distinct. 



Scale of Violence. 



The establishment of a scale of violence is one of the most 

 important steps to be taken at this stage of our studies 

 and information. The question arises at once as to whether 

 this scale should be according to the violence displayed in 

 twisting trees and destroying houses, or according to the 

 property loss. The former would be a very difficult matter 

 to determine, but should certainly be coupled with the ex- 

 lent of region over which the manifestation occurred. It is 

 also true that we should have with every tornado and de- 

 structive storm a definite estimate made, as impartially as 

 possible, of the property loss. We may hope ere long that a 

 beginning of interest will occur and develop in this direc- 

 tion, as has already appeared in the study of fire losses. By 

 what precedes, it is plain that our ultimate arrangement of 

 tornadoes according to a scale would place the more violent 

 in a slightly different position from the more destructive; 

 but this difference is exceedingly slight, and, until we make 

 finer distinctions than are possible now, we can easily place 

 the most violent and destructive in a class by themselves, or 

 on a scale of 3 in class (3). Going to the other end of the 

 scale, we will put the least violent at (1), and (2) will com- 

 prise all between these two. Of course, no hard and fast 

 line can be drawn between these different grades, and 

 some would put a few on the border of (2), in (3), and 

 vice versa, but this is a very small matter. It is very im- 

 portant that in our lists the dates be arranged chronologi- 

 cally; but, if we wish to make a distinction between torna- 

 does at the borderlines, we may do so by using the plus 

 and minus signs : for example, (3 — ) would be near to 

 (2 -\- ), and (3 -|- ) would be the highest in the scale. This 

 would be a step toward dividing into nine classes instead of 

 three. Another method of classification would be to deter- 

 mine the number of houses destroyed in each tornado, and 

 then arrange the list on this basis. Such a list would not 

 differ materially from the former. The only diiReulty would 

 be in assigning the proper number of houses under each 

 scale. For purposes of comparison and in computing risks, 

 we might also arrange a list of towns destroyed, or partly 

 destroyed, according to an arbitrary scale. It matters little 

 what method we adopt, for it is probable that any scheme 



