SCIENCE. 



[Vol. XIII. No. 313 



cloud a little denser and darker than the clouds are generally. 5. 

 A hail-storm is an incipient tornado in the cloud-region of a " low." 

 6. Tornadoes always occur in the south-east quadrant of a " low," 

 and at distances generally of from 200 to 500 miles from the centre. 

 They are not likely, however, to occur in the south-west and north- 

 west quadrants of a "low," because the currents and counter cur- 

 rents there are nearly east and west ; and hence the effect is neu- 

 tralized, tending neither to produce nor destroy the unstable state. 

 As the " low " progresses eastward, the region of country lying, on 

 the average, about 350 miles to the south and east of the centre of 

 the general storm at any time, is the region within which tornadoes 

 may be expected. 



The destructive violence of a tornado is sometimes confined to a 

 path a few yards in width, or it may widen to the extreme limit of 

 eighty rods. The tornado, with hardly an exception, occurs in the 

 afternoon, just after the hottest part of the day. The hours of 

 greatest frequency are 3.30 to 4 P.M., and 4.30 to 5 P.M. The de- 

 structive power of the wind increases rapidly from the circumfer- 

 ence of the storm to its centre. Observations with a single barom- 

 eter will not indicate the approach of a tornado, however near 

 the position of the instrument to the path of the storm ; and such 

 observations are of value in this connection, only when a number 

 of them are displayed upon the weather-map. The tornado season 

 includes the months of March, April, May, June, July, August, and 

 September. There are, however, cases in a long series of years 

 where tornadoes have been reported every month of the year. The 

 months of greatest frequency, as determined from a record of 206 

 years, are April, May, June, and July. The month of greatest fre- 

 quency is May, April coming next on the list. The State in which 

 the greatest number of tornadoes have occurred is Missouri, fol- 

 lowed next in order by Kansas and Georgia. The 425 tornadoes 

 and •' windfalls " recorded in Wisconsin far exceed the number 

 from any other Siate ; but little weight can be given this fact, 

 ■owing to the want of a similar investigation of the subject of 

 " windfalls " in other States. The violence of tornadoes expressed 

 relatively by States places Missouri first, succeeded by Iowa and 

 Alabama. By "violence" in this sense is meant the most com- 

 pletely developed storms with perfect conditions longest sustained. 



Considering the entire record of eighty-eight years (years of rec- 

 ord from 1682 to 1888), nearly 4,000 persons have been reported 

 killed, and 6,000 injured. This record is very imperfect, owing to 

 the large number of cases in which the killed and wounded were 

 not definitely reported. The States in which tornadoes have 

 proved the most destructive to life are in relative order as follows: 

 Missouri, Mississippi, Iowa, Illinois, Minnesota, Wisconsin, and 

 Ohio. 



Considering the reported valuation of property destroyed, the fol- 

 lowing States have experienced the most destructive storms, and 

 in the order named : Missouri, $94,325,000 in forty-seven years ; 

 Ohio, $87,737,500 in eighty- four years; New York, $67,000,000 in 

 -on hundred years ; Kansas, $64,000,000 in twenty-nine years ; 

 Georgia, $56,500,000, in ninety-three years ;• Minnesota, $50,750,000 

 in thirty-three years ; Iowa, $49,575,000 in forty-five years; South 

 ■Carolina, $46,875,000 in one hundred and twenty-seven years. 

 These values are necessarily approximate, owing to the imperfect 

 reports, and it is believed that they fall considerably short of the 

 actual amount of loss. 



It is a difficult matter to obtain reliable estimates of the actual 

 value of property destroyed by tornadoes in the United States, 

 since the date of the earliest records; viz., June, 1682. Although 

 the average yearly occurrence of these storms has probably re- 

 mained unchanged, and will continue so, yet the value of the prop- 

 erty subject to destruction by them has increased enormously, and 

 the amount is constantly enlarging as the country advances in 

 wealth and population. Where fifty years ago the tornado swept 

 .a barren plain or trackless forest, it now crushes and overwhelms 

 prosperous cities, and devastates rich and populous agricultural 

 districts. 



The approximate aggregate loss to property by tornadoes, con- 

 sidering the number of storms reported, is nearly one billion of dol- 

 lars. Considering the number of years embraced in the period of 

 ■observations and the average yearly number of tornadoes, the 

 .amount is nearly ten billions of dollars. This estimate is probably 



in excess of the actual loss, because no allowance is made for the 

 many storms which have occurred since 1682 without appreciable 

 waste. Taking the past thirty vears as affording more satisfactory 

 information, and considering the average yearly occurrence of tor- 

 nadoes as 146, the approximate aggregate property loss foots up 

 about two billions of dollars. 



A comparison of yearly values, as obtained from carefully pre- 

 pared tables showing the number and geographical distribution of 

 tornadoes, might lead one to the conclusion that these storms were 

 on the increase. Such a deduction would certainly be erroneous, 

 and for the following reasons. First, From a careful investigation 

 of the origin of tornadoes and their geographical distribution, there 

 -is every reason to believe that these storms were as frequent and 

 violent two hundred years ago as now. Moreover, there appears 

 to be no cause for any unusual change in the annual frequency of 

 tornadoes for a like period to come. Second, It must be considered 

 that during the past ten years the Signal Service has had great 

 facilities for collecting reports ; and the rapid growth of the 

 country, with a greater zeal of the press, has brought to light the 

 occurrence of many storms which otherwise would not have been 

 reported. Third, The statistical tables are not sufficiently com- 

 plete, especially prior to 1875 (without which period it would not 

 be safe to make deductions), to permit of reliable conclusions as to 

 periods of maximum and minimum occurrence. It is not unlikely 

 that such variations exist, and that they depend upon the relation 

 of heat and moisture to the general condition of the atmosphere. 

 The more frequently the unstable state of the atmosphere is pro- 

 duced, together with a relative gyratory motion, the more favor- 

 able are the conditions for the occurrence of tornadoes, and vice 

 versa. Fourth, In the region between the 95th and 1.07th meridi- 

 ans, tornadoes still occur without causing much damage, because 

 of their passage over thinly settled portions of the country. Owing 

 to this fact, little attention is given to these storms ; but this indif- 

 ference will soon disappear, as the country rapidly settles up and 

 every appearance of the cloud-monster is marked by death and 

 destruction. Fifth, The years (118) that are missing in the period 

 of 206 years from 1682 to 1887 inclusive are not to be considered 

 as years in which no tornadoes occurred, but as years in which rec- 

 ords are wanting owing to failure of observations. 



Considering the past ten years (1878 to 1887 inclusive) as fur- 

 nishing reliable and exhaustive records of tornadoes, and that the 

 period prior to 1878 (196 years, 1682 to 1877 inclusive) is deficient 

 owing to want of facilities in collecting reports, we may give an in- 

 terpolated value for each of these latter years, as determined from 

 the complete ten-year record. This value is found to be 146, 

 which means, that, on the average, 146 tornadoes will occur in the 

 United States. 



Applying this constant to existing records, we have a grand total 

 of storms from 1682 to 1887 inclusive (206 years) of 30,076 torna- 

 does, instead of 2,435 actually observed and reported. This would 

 indicate a failure to report the occurrence of about 27,641 tornadoes 

 which have probably passed over portions' of this country since 

 1682. In that year a very destructive tornado, with distinct funnel- 

 shaped cloud, visited New Haven, Conn., at 2.30 P.M. on the loth 

 of June. It is the very long interval of 1 18 years, during which rec- 

 ords are entirely missing, that makes the discrepancy so great in 

 the grand totals. 



No well-authenticated case of a tornado has been reported from 

 the region of country lying west of the 105th meridian, and it may 

 be generally stated that these storms do not occur in the United 

 States west of the looth meridian. The cause for this is found in 

 the lack of favorable conditions, on account of the dryness and the 

 lower temperature of the air, and the want of uniformity in the 

 direction and force of surface currents. Violent straight winds, 

 attended with considerable destruction to property,, have been re- 

 ported several times in the past fifteen years, from southern and 

 central California, Arizona, New Mexico, and Montana. 



Much has been said and published concerning the influence of 

 forests upon the occurrence and destructiveness of tornadoes, and 

 many people believe that where timber grows in great abundance, 

 tornadoes cannot occur. By comparing the number of tornadoes 

 in each State with the acreage of forests, as estimated in the last 

 census report, it is found that the latter appear to have no percep- 



