ON GEOLOGICAL TIME. 71 



will be calculated just as I have calculated that of 

 the supposed melting of ice. Thus meteors 

 falling on the earth in fine powder (as is in all 

 probability the lot of the greater number that 

 enter the earth's atmosphere and do not escape 

 into external space again) enough to form a layer 

 about T>V of a foot thick in 100 years, if of 2*4 

 times the density of water, would produce the 

 supposed retardation of io s on the time shown by 

 the earth's rotation. But this would also accelerate 

 the moon's mean motion by the same proportional 

 amount ; and therefore a layer of meteor-dust 

 accumulating at the rate of -fa of a foot per 

 century, or I foot in 4,000 years, would suffice to 

 explain Adams and Delaunay's result. I see no 

 other way of directly testing the probable truth of 

 M. Dufour's very interesting hypothesis than to 

 chemically analyze quantities of natural dust 

 taken from any suitable localities (such dust, for 

 instance, as has accumulated in two or three 

 thousand years to depths of many feet over 

 Egyptian, Greek, and Roman monuments). 

 Should a considerable amount of iron with a 

 large proportion of nickel be found or not 

 found, strong evidence for or against the meteoric 

 origin of a sensible part of the dust would be 

 afforded. 



Another source of error in the earth as a time- 

 keeper, which has often been discussed, is its 

 shrinking by cooling. But I find by the estimates 



