ItoRUAKT 21, 1919] 



SCIENCE 



185 



of the air as vertical convection may produce 

 will be essentially along meridians, and thus 

 a function of the time of day. But, in gen- 

 eral, convection increases most rapidly during 

 the forenoon, say eight to nine o'clock, is most 

 active at ten to eleven o'clock, and reaches its 

 greatest elevation about four o'clock in the 

 afternoon. Hence the damming up of the 

 atmosphere, due to vertical convection, and 

 the resulting increase of barometric pressure 

 must increase most rapidly during the fore- 

 noon, and come to a maximum about ten 

 o'clock. After this the convectional interfer- 

 ence decreases, while at the same time the 

 amount of air in a vertical column of fixed 

 cross-section diminishes as a result of ex- 

 pansion and overflow, until at about four 

 o'clock in the afternoon the barometric pres- 

 sure has, as a result of this overflow, reached 

 a minimvun. 



To form some idea of the magnitude of the 

 barometric change due to convectional turbu- 

 lence, consider the atmosphere between two 

 parallels of latitude near the equator. This 

 limited quantity may be regarded as a stream 

 flowing around the earth, having its minimum 

 velocity and maximum depth where convec- 

 tional interference is greatest, and maximum 

 velocity with minimum depth where convec- 

 tion is absent. And since the linear velocity 

 of a point on the equator is approximately 28 

 kilometers per minute, while during the fore- 

 noon the rate of increase of the barometric 

 pressure at the same place is, roughly, 0.2 mm. 

 per hour, it follows that a damming up, or 

 check in the flow, of the given stream at the 

 rate of 0.44 kilometers per hour would be 

 sufficient of itself to account for the observed 

 rise in the barometer. But if the average 

 velocity of the wind, or flow of the stream in 

 question, is 10 meters per second, which it 

 may well be, the rate of decrease in velocity 

 requisite for the given rate of pressure in- 

 crease could be produced by having only 1 

 part in 80 of the whole superincumbent at- 

 mosphere brought to rest per hour, or tlin 

 equivalent thereof, an amount that perhaps is 

 reasonable. At any rate, the assumed velocity 

 decrease is of the same order of magnitude as 



that observed to take place during, and aa the 

 result of, diurnal convection. 



It apijears then : 



(a) That the afternoon minimum is caused 

 essentially by overflow from the region where 

 the atmosphere is warmest, or better, perhaps, 

 from the meridian along which the tempera- 

 ture increase has been greatest, toward that 

 meridian along which there has been the great- 

 est decrease in temperature. 



(&) That vertical convection interferes with 

 the free horizontal flow of the atmosphere 

 and to that extent dams it up and corre- 

 spondingly increases the barometric pressure; 

 also, that the time of this interference agrees 

 with the forenoon changes of the barometer, 

 and tliat its magnitude is of about the proper 

 order to account for the forenoon barometric 

 maximum. 



The afternoon barometric minimum and the 

 forenoon maximum, therefore, are to be re- 

 garded as effects of temperature increase; the 

 minimum as due to expansion and consequent 

 overflow; the maximum as caused by vertical 

 convection and consequent interference with 

 the free circulation of the atmosphere. 



The forced afternoon minimum would oc- 

 cur in an otherwise stagnant atmosphere, and 

 substantially as at present; but not so with 

 the forced forenoon maximimi, since the inter- 

 ference or damming effect depends upon a flow 

 or circulation of the atmosphere, parallel 

 roughly to the equator. 



It remains now to accoimt for the night 

 ten o'clock maximum and four o'clock min- 

 imum, both of which appear to depend upon 

 the natural or free vibration of the atmos- 

 phere as a whole. 



This subject has been discussed by several 

 mathematical physicists of great eminence. 

 The latest and most complete of these dis- 

 cussions is by Lamb, who concludes : 



Without pressing too far conclusions based on 

 the hypothesis of an !Xtmo3phere uniform over the 

 earth, and approximately in convective equilibrium, 

 we may, I think, at least assert the existence of a 

 free oscillation of the earth's atmosphere, of 

 " semidiunial' ' type, with a period not very dif- 



