Appendix B. 



In fact, the temperature l)eing 23° '9 Celsius (75" F.) at Fort Portal, ami 

 r-6 Celsius (34°-4 F.) at Edward Peak, a difference of 22°-3 Celsius (72°-6 F.), 

 if the decrease occurred proportionately to the altitude the temperature of 

 Bujongolo should be aliout 7°-2 Celsius (-to' F.), whereas it is only 3°'9 Celsius 

 (39°7 F.). This, therefore, means that the column of air has a lower temperature* 

 than is assumed liy the theory, so that to an equal difference of pressure 

 corresponds a less difference of altitude. 



Lastly, in connection with determinations of this nature, it is not to be 

 forgotten that results now well established are : — 



1. That heights calculated by means of thermo-barometric observations 

 are generally found to be greater with measurements made by day compared 

 with those made by night. They present a considerable daily range with the 

 maximum value shortly before the maximiuii daily temperature, and the 

 minimum one or two hours before sunrise. The extent of range is influenced 

 bj"- the season, the local conditions, and the state of the sky. 



2. That the altitudes calculated with the niean daily or monthly values of 

 the observations are found to be too low in winter and too high in summer ; 

 with the annual mean they differ little from the actual heights. 



* Caused probably by the great masses of ice on Buweiizori. 

 370 



