in @ Voyage across the Atlantic. 119 
ceptible of being rendered one of great delicacy in a very sim- 
ple form. 
As degrees of dryness, referred to the varying temperature 
of the atmosphere, want a common measure for purposes of 
comparison; to obtain one, I deduce from the observed tem- 
peratures of air and vapour, the actual capacity of the atmo- 
sphere for moisture and the quantity present: the expression 
of the ratio of that capacity to the difference, gives the dryness 
or relative quantity deficient to saturation. It is expressed in 
centesimals. Its complement is the proportion of moisture 
present at the time of observation. 
The same purpose would be answered by computing the 
elastic force of vapour from the temperature and degree of 
dryness. 
In employing a thermometer for its usual and direct purpose, 
precaution is requisite to avoid error in observations taken on 
ship-board ; for it is the temperature of the sea or of the atmo- 
sphere, not that of the ship, which is to be ascertained. It 
was my practice to expose the thermometer over the windward 
side of the vessel, or to suspend it out of the cabin-port or 
stern-window, in the shade, and in a situation to have free 
access of air, unaffected by radiation or reflection from any 
thing appertaining to the ship. 
In using the thermometer for hygrometry, similar care was 
employed to present the instrument to the free access of air 
from the sea, independent of the influence of a wet deck, or 
any other presumable source of error. Many of my earlier 
observations have been rejected, as, for want of some precau- 
tions, they are liable to suspicion of error. 
To try the temperature of the sea’s surface, a thermometer 
was sometimes let down to it. More frequently sea-water, 
fresh drawn, was examined, 
In general the temperature of the sea’s surface differs little 
from that of the air immediately over it; or the small differ- 
ence, which is perceivable at one time of the day, is reduced at 
another; when the degree of heat exhibited becomes nearly 
