u 



KNOWLEDGE. 



[Februaby 1, 1900. 



and me/i in black coats, their sleeves flying out behind 

 them, liiiiTy into the nearest villages from the blank 

 and stricken fields. TJ]) in the Alps you may ';ioucli 

 beneath a rock and see tlie lightning leap from scar to 

 scar, and hear the church bells ringing out their answer 

 from the fields a thousand feet below ; but here, in the 

 great open, the elemental forces will hold you more se- 

 ciu'cly in their grasp. A hand, as it were, at first 

 gentle, then imperious, pushes you forward from be- 

 hind. In the air there are resistless spirits ; in the 

 oak-forest there are strange whisperings, and the cry of 

 frightened birds. The herd-boy, with his rustic pipe, 

 knows these things better than the wisest student of 

 geography. 



LONG WAVES OF WINTER WEATHER. 



By Alex. B. MacDowall, m.a. 



The student of weather uses various time-units ; hours, 

 days, months, seasons, years, &c. Each of these is a 

 groui:)ing of smaller units ; and proceeding further, he 

 may compare groups of five years, or ten years, or more. 

 This is often instructive; and it is still better, in the 

 case of a given year-group, to compare, not merely suc- 

 cessive groups of that order, but overlapping gi'oups, 

 e.g., in the case of ten-year gi'oups, not merely 1841-50, 

 1851-60, etc., but 1841-50, 1842-51, 1843-52, etc. Here 

 we come to the principle on which cui-ves are often 

 smoothed. Instead of observing how a given element 

 varies from year to year, we note how the average of 

 three, of five, of ten years (or more) varies from year 

 to year. (The principle obviously applies to smaller 

 units also.) 



Some interesting relations are thus brought out; and 

 light may be thrown on those longer waves of variation, 

 which often underlie, and are rather obscured by the 

 minor undulations. 



In this jiaper, I pi'opose to look at winters in groups 

 of ten. The term " winter " will be used somewhat 

 loosely, and winters may be briefly denoted by the year 

 in which they end (1842, e.g., meaning 1841-42). 



Let us begin with wind in the winter-half (October 

 to Mai'ch). Taking the Greenwich tables of wind- 

 distribution in which all winds ai'o reduced to the four 

 cardinal directions, we may ask, How many days of 

 northerly and easterly wind (the two colder directions 

 in winter), occurred in each of the winters from 1860? 

 The reply apj^ears in the zigzag curve A. This is tra- 

 versed by a thick line curve, each year-point of which 

 represents an avei-age of ten; e.g., the first, 1865, the 

 average ot 1861-70; the second in 1866, that of 1862-71, 

 and so on. This average, it will be seen, rises on the 

 whole, from a minimum in 1870 to p. high point in 1890. 

 That is to say, in the ten winters about 1870, little 

 N. and E. wind; in those about 1890, much. 



Next, as to temperature. B is a curve derived from 

 that of the mean temperature of the group of four 

 months December to March, since 1842, by the same 

 averaging process. It is an inverted curve ; the high 

 points meaning low temperatures, and the low ones 

 high. Here we find two conspicuous wave crests 1856 

 and 1890, while a minimum (maximum of mildness) 

 appears at 1873. 



Thirdly, rainfall. In the coldest parts of the year, 

 severe cold and dryness are generally associated, while 

 great mildness often goes with wetness. Treating the 

 rainfall of October to Mai'ch in the same way, we get 

 the (inverted) curve C ; and it is not surprising to find 

 a gen'jral similarity to that of temiaerature. The ten- 



winter groups about 1856 and 1890 ('88 the highest 

 point), show a deficiency of rainfall ; those about '73 

 ('72) an excess. 



jWW '-f '"« S2 'b' '60 V » '72 '6 W '4 '8 92 '6 1300 



A. Days of N. + K. wind in October- ]y[arch (Gr.), emoothoJ with 

 areniges of 10. B. Mean T. (Orr.) in December-March, smoothed. 



C. Kainfall, October - March (Gr.), smootlied. D. Frost dajs, 

 September- May (Gr.), smoothed. E and C are inverted. 



Lastly, frost days in September to May. The cui've 



D, obtained in the same way, agrees with B in its 

 maxima (of cold) at '56 and '90; but the dip between 

 reaches its lowest point a little earlier, in '63 or '65. 



We have, then, the outstanding fact, of a great peak, 

 or wave crest, of cold about 1890 ; the ten-winter group 

 1886-95, being the coldest of all the 49 groups con- 

 sidered. That group is also conspicuous for its quantity 

 of N. and E. wind, and shows less than the average 

 precipitation. It may be useful to look at those ten 

 winters, from the standpoint of frost days. The 

 average in September to May being about 55, wo have 

 the following numbers and relations: — 



'86. '87. '88. '89. '»). '91. '92. "93. '94. '95. 

 Frost davs ... 75 80 90 CO 45 82 73 48 42 62 



Kelatioirto average -f 20 -f 25 -H 35 -h 5 - 10 + 27 -f 18 -7-13-1-7 



Thus we see that seven out of those ten winters were 

 severe; that 1888 had the largest number of frost 

 days; and that 1890 itself was one of the three mild 



