of the Island of Barbadoes. 63 



The means of temperature for 9 a. m., 3 p. m., and 9 p. m., in the above 

 table, for nine months, indicate a curve very closely resembling that 

 derived from the observations in June, July, and August. The differ- 

 ences between the temperatures at these hours, for the whole period, 

 are rather less than between those for June, &c. ; but as the mean diur- 

 nal variation, as shewn by the register thermometers, is less in the former 

 than in the latter case, in almost exactly the same proportion, the dimi- 

 nution of the differences between the temperatures at the respective hours, 

 the consequence of a somewhat flatter curve, was to be expected. The 9 

 A. M. observation, for the whole period, approaches that of 3 p. m. more 

 than, according to the hourly observations, it should do ; but when it is 

 recollected, that in the months given in the table, almost the whole of 

 those in which the equation of time is additive to mean time, are in- 

 cluded, and in which, consequently, the sun's hour angle is less, and 

 this heating power in the forenoon greater than indicated by mean time, 

 while February and March, during which the equation of time is sub- 

 tractive, and reaches its greatest amount, are omitted, there is good 

 reason to suppose, that, had the observations been continued for twelve 

 months, the results obtained would have presented a perfectly satis- 

 factory agreement with those afforded by the hourly observations. 

 These results will, in the absence of more extensive materials, afford 

 data for arriving at a very close approximation to the annual mean 

 temperature of Barbadoes, from any series of observations, and they 

 will, in all probability, be equally applicable to all the lesser West India 

 Islands, and to the flat portions of the adjacent coast of South America, 

 where the free course of the trade winds is not interrupted by moun- 

 tains. 



The dew point in the table is calculated from the indications of a 

 wet bulb thermometer ; to show how far these could be depended on, I 

 frequently determined the dew point directly, in September and Octo- 

 ber, and again in December and January'', by cooling a small quantity 

 of water in a piece of common test tube, by powdered nitre, the tem- 

 perature of the mixture being obtained, at the moment of deposition of 

 dew on the surface of the tube, by a delicate thermometer, with which 

 the mixture was constantly stirred until the deposition took place. 

 This method, with a little management, is capable of great precision. 



During September and October 31, observations of the dew point were 

 obtained at 9 a. m., and 28th at 3 p. m. These, and the corresponding 

 results from the wet bulb thermometer, at the same hours, and on the 

 same days, are as under — 



9 A. M. 3 p. M. 



Dew point calculated, . . . 73°.44 73°.63 



„ observed, . . . 73 .04 73 .24 



Actual calculated in excess of the dew point, 0°.40 0**.39 



