T5 



absence of fires during tlie inontlis of July ami Aui;ciist, Imt from Se[itciiiber, 

 tliey increase in number s^rowing greater and greater until a niaxinmm is reached 

 in tlie month of ilarch ; then the number gradually decreases till it becomes zero 

 in July. Tliis fact is more strikingly sliown in Chart A in which the upper 

 brokea line indicates the number of fires in different months, the scale of nimiber 

 (if tires being on the upper half of the left edge of the chart. The chart tells its 

 own story too well to need any comment. The lower broken line represents the 

 number of miles of wind during each month, the scale of the number of miles 

 being written on the right lower edgi' of the chart. The number of miles is the 

 average from observations during the years ISTi) and ISSO, so that the broken 

 line or the curve may not exacthj represent the true average; I am, however, of 

 oi)inion, that it is sufficiently near the true average for our purpose. In com[Kiring 

 these two curves, one is struck with a remarkable general similarity between the 

 two, showing tlie obvious truth, that the wind is the main cause of great fires. 

 In botli these curves, a maximum occurs in tlie month of March. Tlie fact that 

 March is the windiest month in year, no doubt, makes the freipieucy of fires 

 during that month the greatest. One important fact noticeable in the chart, is 

 the remarkable difference between the two curves as regards the month of July, 

 wliich can be explained by the following consideration : although winds are more 

 frequent in July than in the adjacent months, as «e have little need of fires 

 (hiring the hot summer months, the danger of a fire originating accidentall)' is 

 greatly lessened and the probability of a gre.it fire at this time is proportionately 

 diminislicd. This is undoubtedly the cause of the difference between the two 

 curves. The curves in Chart B, are very similir to those in Chart A ; except that 

 the lower curve is drawn so as to show the total number of high winds during 

 the years 1879 & 18S0 (in the chart, the heading ought to read "tot^d number 

 during 1S79 & 1880" instead of "mean of 1870 & ISSU)". In their general 

 characteristics the two curves agree remarkably well except in llic iininlli ul 

 August, which can bo explained by the same consideiation as before. As an 

 additional explanation of the absence of great fires during .July, the lower curve 

 in this chart shows a frreat depression for July indicating that there is a very 

 small number of liigh winds during this month, and consequently a smaller [iro- 

 bability of the occurrence of great fires. In Chart E, the liorizontal straight lines 

 arc drawn so as to nhow tlie relative magnitudes ef all the fires in Table A ; in 

 each, a line representing a fire is drawn so as to fall on a ]iarticular point in the 

 niontli, representing (lie day <ii occurivncc and lieitJ again lines are crowded 

 tiigetlier in the monlli nf March. Charts C and Dare very similar to Charts A 

 anil 15: Imt instead «if lieing referred to time tli(^y are referred to directions. 

 With reference lo directions, in Tables B and C, we see at a glance, tliat nearly 

 all the fires liavo tlu! directions N, NW, S and SW ; and that those from {\h\ N 

 and NW are tlic greatest in niunber. lleferring to Chart C, we begin at 10 with 

 no fire, one willi SK, and S ami SW have res|H-clivcIy 10 and i:>. The 

 curve falls III zero at W; then ijscs inoruioiislv at NW and iilfaiim a nia\i- 



