188 GENERATJZATION OF THE PRECEDING TABULAR RESULTS 



stations with series of observations shorter than four years the position was marked 

 as before, but only the nearest wliole inch was written against the dot ; the relative 

 value of the results was thus, in a measure, indicated in the construction of curves 

 of equal rain-fall. These curves were drawn with a free hand among the dots by 

 graphical interpolation, and with due regard to the importance of long and short 

 series. These curves, designated as iso-hyetal lines, and constructed in the manner 

 of contour lines generally, are graduated for certain equal increments of rain, the 

 diiference between adjacent curves resulted from a consideration of the probable 

 uncertainty of the results. If drawn too close, that is if too many curves were 

 shown, they would exhibit temporary or accidental inflections, which could only tend 

 to complexity and confusion ; on the other hand, if the curves were too wide apart, 

 there would be danger of losing portions of permanent features in the distribution. 

 The distinction between long and short series in the graphical process is of import- 

 ance in a phenomenon of such great variations from year to year and from the same 

 season in different years, and the numbers of the second class given in whole inches 

 were used to complete, modify, and generally to improve the curves resting upon the 

 more reliable data. Special consideration was necessary to select for each chart those 

 particular curves and their graduation which would best bring out its leading 

 features, and further, to facilitate the undei'standing and the ready interpretation 

 and use of the charts color-shading was introduced. The curves, whether principal 

 or intermediate, are indexed, and can thus be easily followed by the eye, and each 

 chart is supplied with a sufficient explanation to be understood, even when detached 

 from the text. It was neither necessary nor practicable, from want of space, to indi- 

 cate on the charts the individual stations and their amount of rain-fall, though they 

 are crowded in on the manuscript charts. Tlius, the number of stations plotted and 

 utilized for the chart of the mean annual distribution is about 1,300, and the numbers 

 are larger for each of the season charts. Comparing the new with the old rain charts, 

 the superiority of the former will be apparent, and while perhaps too much detail was 

 given on the charts of the first edition, which, through the increase of observations, is 

 now known not to form part of permanent features, but arose from insufficient data at 

 that time, yet the apparent distortion of the curves, when the two sets are compared, 

 may be produced by small changes in the amount of rain-fall, and, while the general 

 features are preserved, the present charts will bring them out only more prominently 

 and broadly. We now pass to the special consideration of each of the charts. 



GENERAL CHARACTER OF THE DISTRIBUTION OF THE RAIN-FALL ON THE AVERAGE 



DURING THE YEAR. 



A knowledge of the distribution of rain over the country is most readily obtained 

 by inspection of the fifth chart of the series, viz., that showing the average annual 



