TEMPERATURE VARIATIONS. 285 



tures, even though these fractures be too small to be detected by the 

 naked eye. Such cracks are therefore the lines of least resistance along 

 which decay takes place, and once they are opened the work of destruc- 

 tion is begun and will never cease until the rock is completely decom- 

 posed. It is for this reason that I lay no special stress upon the frequency 

 of changes. Such changes would, however, be of the utmost importance 

 if they sank below and rose above the freezing point. This, however, 

 does not occur in Brazil. 



Temperatures in the sun's rays cannot be adduced from temperatures 

 in the shade, and observations made in the sun's rays, even at the same 

 time and place, vary greatly according to the instruments used and 

 the methods of exposure. Without direct observations we can therefore 

 get only an approximate idea of the changes of temperature suffered by 

 the openly exposed rocks. The total absence of such direct observations 

 is the only excuse for the approximations given below. 



In 1876, in the dry interior of Pernambuco near Aguas Bellas, I made 

 some observations to ascertain the difference between the temperature 

 in the sun and in the shade. The weather was sultry, and at the hottest 

 part of the day — between 2 and 2.30 p. m. — the thermometer in the sun, 

 and covered with one thickness of black woolen cloth, registered respect- 

 ively 40, 48, and 50 degrees higher in the sun than in the shade on the 

 three days on which the observations * were made. 



If we assume from this suggestion that the temperature in the sun 

 during the hottest part of the day is 50 degrees, Fahrenheit, higher than 

 it is in the shade, we may use the shade readings to determine the max- 

 imum sun reading as given for some of the stations mentioned in the 

 table on page 286. 



These observations are, to be sure, very desultory and incomplete at 

 best, but they are the only ones now available. I cannot refrain from 

 expressing doubt about the total range of temperature suggested by 

 these figures. My own observations were made during the cooler part 

 of the year in all cases. Although impressions are not to be trusted for 

 such purposes, I am confident that properly conducted records will show 

 that exposed rocks are subjected, except along the coast, to a change of 

 temperature of considerably more than 100 degrees, Fahrenheit. 



We are obliged, however, to confine ourselves, for the present, to the 

 data afforded by this table. 



Temperature and exfoliation. — The application of the laws of expansion 

 and contraction of rock masses to the Brazilian gneisses would be simple 

 enough if we had to deal with uniform temperatures ; but the tempera- 



* The thermometer used for these observations was an ordinary eight-inoh instrument mounted 

 on a black tin frame with a metal back. I have no doubt it required correction, but it was bor- 

 rowed for the occasion, and the errors could not be determined and corrected. 



