WEATHER AND CLIMATE 141 



perature 1 . So generally is it true that heating a body causes 

 it to increase in size, and cooling it causes shrinkage in 

 volume, that any exception might be wholly disregarded. 

 But the peculiar behavior of water must not pass unnoticed. 

 As water is cooled more and more it shrinks in its volume, 

 as might be expected, till a temperature of about 4 C. is 

 reached. Further cooling results in its expanding till it 

 changes to ice. As previously noted there are many in- 

 stances of increase in volume where change from liquid state 

 to solid occurs, but water when cooled begins to expand while 

 yet several degrees above the freezing temperature. This 

 makes 4 C. the temperature of maximum density for water, 

 and any change from that temperature, whether rise or fall, 

 involves increase in its volume. 



The amount of expansion (and of contraction) is relatively 

 small in solids, is more in liquids, and is relatively large in 

 gases. Every substance has its own so-called coefficient of, 

 expansion. This fractional number is found by dividing the 

 increase (or decrease) in size for one degree temperature 

 change by the size before the change in temperature was 

 made. In dealing with gases the pressure under which they 

 are confined must be kept constant, and the volume of the 

 gas at the freezing temperature of water (o C.) is the basis 

 for comparison (division). 



The coefficient of expansion for mercury is sufficiently 

 small so that in the large tubes of barometers any rise and 

 fall of the mercury column by reason of temperature changes 

 of several degrees may be considered a negligible value for 

 all general purposes. Thermometer tubes, however, have 

 an exceedingly small bore. The difference in levels of the 



1 For the heat of vaporization and of fusion, where changes of solids and 

 liquids and gases from one state to another are considered, see page 95. 

 Neither change in volume by reason of change of state, nor consideration of 

 any heat liberated or made latent in any such changes, is involved in studies 

 of the thermometer. 



