42 MEMOIRS OF THE NATIONAL ACADEMY OP SCIENCES. 



the shade of a high cocoanut tree, there being several other trees in the immediate vicinity. The 

 roof sloped towards the north, and after April 29 was surmounted by an additional roof raised 

 about 4 inches above the first. The north side of the shelter contained a door, which was fre- 

 quently open. This arrangement of the shelter proved very satisfactory. The whole shelter was 

 shielded from the direct rays of the sun, except for a short time soon after sunrise, by the branches 

 of the trees, and the rays which came through these branches during the heated portion of the day 

 fell upon the upper roof. The important condition of a free circulation of air about the instruments 

 was also secured, since the branches of the trees were from 10 to 20 feet above the shelter, allow- 

 ing the wind to blow beneath them. Three sides of the shelter were made of open louvre work, 

 Mhile the fourth was wholly open. In order to test, by actual observation, the success of the above 

 method of ex^iosure, a series of readings was made of the dry-bulb thermometer, and of a thermom- 

 eter whirled rapidly in the open air. The readings were made under various conditions during sev- 

 eral days, and are given at the end of this report. They show that the readings of the dry-bulb 

 thermometer may be accepted without correction as giving closely the temperature of the air. 



Within this shelter were placed the maximum, minimum, dry-bulb, and wetbulb thermometers, 

 and beneath it, one foot above the grass, a thermometer adapted for measuring the minimum tem- 

 perature at night close to the ground. The bulbs within the shelter were elevated 5 feet above 

 the ground. The aneroid barometer was of the Goldschmid pattern, and proved to be a good 

 instrument. It was generally kept in the house, but during the frequent observations on April 29 

 and May 6, was for convenience placed in the shade in the open air or in the instrument shelter. 

 Comparisons between the readings of this barometer and the mercurial barometer furnished Mr. 

 Preston by the United States Coast and Geodetic Survey were frequently made, and are given 

 at the end of this report. From them a constant correction of + 0.104 inch to the aneroid baro- 

 meter was derived, and has been used in the reduction of the readings. The elevation of the in- 

 strument was about 10 feet above the level of the sea. 



The anemometer was mounted on an iron rod, 9 feet 7 inches above the ground, in a large 

 open space to which the wind had good access, especially from the north and east, the only direc- 

 tion in which the wind was observed to blow during the period of observation. 



The rain-band spectroscope was read occasionally only. Its readings at the different observa- 

 tions were nearly the same, as will be seen from the record given at the end of this report. In 

 damp insular positions in the tropics the instrument seems to give about the same indications from 

 time to time, and therefore is of less value as a hygrometer than in the temperate zone. 



The instruments were read daily at frequent intervals between 6 a. m. and 9 \>. m. As the 

 principal object of the observations was to furnish a basis of comparison for those made in connec- 

 tion with the eclipse, the record is especially full between the hours of 10 a. m. and 2 ]>. m. The 

 observations are given in detail at the end of this report. From them may be gleaned the follow- 

 ing facts bearing upon the meteorology of the island. 



GENERAL WEATHER CONDITIONS. 



With the exception of one day, May 4, the conditions were uniform, and characterized by lair 

 weather with cumulus clouds passing rapidly over the sky, giving rise to frequent but slight 

 showers. Upwards of twenty of tliese showers were noted during the two weeks, the rainfall at 

 each being small. The sky was sometimes clear for several hours, especially in the middle of the 

 day and in the early evening. The clouds observed were almost wholly of the cumulus class, 

 cirro-stratus being also seen occasionally. The only weather disturbance occurred on May 4, 



