CEYLON BRANCH — ROYAL ASIATIC SOCIETY. 169 



In the diagrams appended to this, I have projected the sa- 

 natory phenomena of several individual years, and also those 

 which result from taking the average of all to which my data 

 extend. The plan of these diagrams is simply this ; — the hori- 

 zontal lines denote days of sickness due to a hundred men, and 

 are narked from ten to ten days, the vertical lines (not the 

 spaces between) denote the several months. The results appear 

 to me very interesting. To revert to that which shows the aver- 

 age of eight years, we find 



That from January to May the health of the station gra- 

 dually improves, that it decreases during June, July and 

 August ; improves during September and October, the latter be- 

 ing the healthiest month in the year, and then rises rapidly to 

 December which is the worst, whence it again falls as be- 

 fore ; that is, this table also shows four distinctly marked 

 seasons. 



These seasons however are not synchronic with those indi- 

 cated by the greater or less quantity of rain, thus, the great 

 dry weather extends from May to October, the health of the 

 station is however bad from June to August, and rapidly im- 

 proves in September and October. 



The great rains extend from the middle of October to De- 

 cember, the first of these months is the most healthy, the two 

 latter are among the least so of the whole year. 



The lesser dry weather extends from January to the middle 

 of March, and the latter rains from that time till May, and it 

 appears that during the former period the station is less healthy 

 than during the latter; I have attempted to show the relative 

 positions of these seasons in the annexed sheet. The most strik- 

 ing features in this comparison, are first, the correspondence in 

 number, secondly the want of correspondence in time ; and thirdly 

 what appears least explicable, a want of apparent rule in 

 these variations. A more careful examination of the subject 

 however explains, I think, this anomaly: thus we find that du- 



