Febkuaby 18, 1898.] 



SCIENCE. 



243 



College as lecturer on the application of statis- 

 tics to social and political science ; George P. 

 McKee has been appointed instructor in physics. 



Pkofessoe Chaeles R. Richards, director 

 of the manual training department of the Pratt 

 Institute, Brooklyn, has been appointed to the 

 chair of manual training in the Teachers' Col- 

 lege, Columbia University. 



At the University of Cambridge, Mr. F. C. 

 Kempson and Mr. R. H. Biffen, of Gonville 

 and Caius College, have been appointed dem- 

 onstrators of anatomy and botany respectively. 



DISCUSSION AND CORRESPONDENCE. 



WEATHER HARMONICS. 



The study of weather periodicity has, from 

 the beginning of meteorology, attracted, more 

 or less, the time and attention of students. 

 Yet, so baffling and uncertain are the results 

 so far produced that many have been led into 

 the scepticism voiced by a recent writer, who 

 remarks, ' There is, apparently, no periodicity 

 in the recurrence of weather.' It seems to me, 

 however, that this attitude is the same as that 

 of a student who visited the track of a tornado, 

 expecting to find the trees and other debris 

 lying in perfect circles, but on finding the 

 fallen trees lying over each other pointing in 

 different directions, and other debris in tangled 

 confusion, came back and announced his con- 

 viction that no whirl existed in the tornado 

 funnel. In other words, my study of the sub- 

 ject for many years convinces me that it is the 

 complexity of the data, not the absence of the 

 phenomenon, which has induced this scepticism 

 in regard to weather periodicity. 



I am led to the conclusion, which is extremely 

 important if true, that one of the complexities 

 which has helped to obscure weather periodicity 

 is the existence of what may, perhaps, be called 

 weather harmonics, on account of the resem- 

 blance to harmonics in sound — that is, the exist- 

 ence of other periods related to the primary as 2, 

 3, 4, etc. In what follows I shall briefly outline 

 the evidence on which this conclusion is based. 



For the first examples I take the best known 

 and only generally accepted cycles, the annual 

 and daily periods. The first harmonic periods 

 I wish to point out are multiples of a year, 



namely, two, three, four and eight years in 

 length, all of which are continuously acting, 

 but now and then one becomes predominant, so 

 that it may be selected for illustration. 



Thus, over the interior of the United States 

 there were for many years very marked oscilla- 

 tions of pressure, temperature and humidity 

 covering a period of about two years. These 

 were discussed in the American Meteorological 

 Joiirnal, Vol. I., pp. 130 and 528. The data 

 appeared at first to indicate a period about a 

 month longer than two years, but later investi- 

 gation indicates that it is more exactly two 

 years. Three and four-year multiples have not 

 been marked in the United States, but an eight- 

 year period has been well marked. Thus the 

 Chief of the Weather Bureau gives, in his latest 

 report (1897, p. 23), the years of widespread 

 drought in the United States during the last 

 forty years as follows, 1860, 1863, 1870-71, 

 1881, 1887 and 1894-95. An eight-year series, 

 running as follows, 1863, 1871, 1879, 1887 and 

 1895, takes in four out of six droughts. This 

 seems to have been acting with the eleven-year 

 or sun-spot period, the maxima of which occurred 

 about 1860, 1870, 1883 and 1894, and are ap- 

 parently connected with droughts in the United 

 States. In the British Isles during the last 

 50 years three, four and eight-year periods 

 appear to have been equally active, hence no 

 simple rhythm can be selected for illustration. 

 But I desire to call attention to one striking 

 fact. It is well known that harmonic sound 

 waves, after a certain number of oscillations, 

 occur with their like phases together, and form 

 beats, and it might be expected that harmonic 

 weather periods, if they exist, would likewise 

 form beats. Since 24 is a common multiple of 

 2, 3, 4 and 8, extremes of weather would be 

 expected to be separated by such an interval. 

 Now, it is a curious fact that the curves pub- 

 lished by A. B. MacDowall, showing the num- 

 ber of frost days at Greenwich, show very 

 marked extremes at this interval. For exam- 

 ple, the greatest number of frost days were in 

 1855 and 1879, 24 years apart, while the least 

 number were in 1872, 1884 and 1896, separated 

 by intervals of 12 and 24 years. (See Meteoro- 

 logical Zeitschrift, 1897, p. 384. ) 



I have reason to believe there are also periods 



