660 LOWELL—MARS ON GLACIAL EPOCHS. [Nov. 16, 
both, but must consist in some peculiarity of the southern cap. 
Furthermore it must be some very general condition, for the eccen- 
tring is not confined to any one era or stage in the existence of the 
cap. It is observable from the time the cap is at its widest to the 
supreme moment of its decay. Nor does it shift its place to any 
extent throughout the shrinking, till the cap’s size becomes so small 
that a trifling preponderance here or there in longevity unduly ex- 
presses itself in lorgitude. Such constancy of position shows that 
the whole accumulation is eccentrically placed, and not simply that 
some part is so locally conditioned as to outlast its fellows. 
25. To account for the phenomenon, analogy tempts us to jump 
to the conclusion that elevation is responsible for the survivorship. 
And so it has been thought to be by many who have philosophized 
on the subject. For this would be the fact on earth, and the same 
we are, therefore, prone to impute to Mars. But consideration 
shows that such cannot be the case. 
Cold increases'with ascent above sea-level because the enveloping 
blanket of air or rather of water vapor thins out as we rise. On 
thé earth in latitude 45° an elevation of a couple of miles is usually 
enough to bring one into the region of perpetual snow. But on 
Mars this would not be the case. On Mars the cold could not in- 
crease thus with the speed it does on earth. It is possible to affirm 
this without any regard to the actual amount of atmosphere upon 
that planet. For it is a simple matter of physics with which we 
have to do. 
26. The mere mass of a planet decides the distribution of its at- 
mospheric envelope. It does this irrespective of what the amount of 
that envelope may be. No matter how dense or how rare the air be 
at the planet’s surface, the air diminishes upward by a law which 
depends directly and primarily upon the planet’s mass. This law 
is found as follows: 
Since the density of the air varies primarily as the pressure put 
upon it, we have at the point, if D denotes the density, / the pres- 
sure and g the force of gravity 
dp caD = —ap=agD 
“ax ax 
whence 
@D = agdx 
