THERMOMETRICAL OBSERVATIONS. 2^7 



therefore neceflary that the water, or any other fluid 

 made ufe of fhould be entirely free from any undulation 

 both fore, and afternoon, when the obfervations are made, 

 which will not be the cafe if the furface is expofed even 

 to a very light breeze, to effect this purpofe I have made 

 ufe of the following apparatus, viz. 



Plate V. Fig. i. reprefents a tin cup, about 2 inches 

 deep, 5 inches long, and 3 wide ; it is well to have the 

 bottom made heavy by fitting fome lead in it. This cup 

 is to be filled with water and the wind kept from it 

 by covering it with the roof (Fig. 2.) the ends, and 

 lower parts of which are made of tin, and the principal 

 part of the fides, or inclined planes are of talc or ifinglafs ; 

 which fhould be of a good quality, and rendered fuf- 

 ficiently thin by feparating, and taking oft' a number of 

 laminae with the point of a penknife. The lower part of 

 the roof fhould be fo conftrudted, as to go down into 

 the cup about 3 tenths of an inch. The degree of incli- 

 nation of the planes, forming the two fides of the roof is 

 of little importance. The planes of the one I have al- 

 ways ufed ftand nearly at right angles to each other. The 

 lower part of the roof fhould go eafily into the cup, 

 becaufe it fometimes happens that the evaporation from 

 the water, will be fo abundant as to cover the ifinglafs, 

 and render the image of the fun which is reflected from 

 the water obfcure : In that cafe the roof mud be removed 

 a few feconds of time, and the particles of water on the 

 ifinglafs will difappear. As the ifinglafs when properly 

 reduced will be very thin, and conlequently tender and 

 delicate, it is necefl'iry that it ihould be defended againfl: 

 accidents when not in ufe, for this purpofe a cafe of tin 

 fuch as reprefented by Fig. 3. will be found convenient. 

 The equal altitudes in the following work, with a few 

 exceptions, were taken with fextants, fometimes by three 

 perfons following each other as quick as pofl^ble, the 



E e 2 correfponding 



