December 2, 1920] 



NATURE 



437 



plicated. It is possible that this conception of the 

 slow, gradual ascent of air may have a bearing upon 

 the cloud-formation associated with a coming cyclone, 

 but the subject is too long for a letter. 



The fourth aspect is the behaviour of the convected 

 air with regard to its environment. The slowness of 

 its rate of ascent is dependent largily upon the 

 development of eddies and consequent dilution of its 

 mass with the cooler environment. This cannot of 

 itself arrest the upward motion, though it delays it, 

 and, consequently, when the convecttd air has arrived 

 at its ultimate level it will have carried with it some 

 of the air which formed its environment on the way. 

 Hence the rising air will have "evicted " a certain 

 amount of air by its passage. 



The importance of combining these aspects is at 

 once apparent if we consider that convection in still 

 air would simply mean a readjustment of the mass 

 in the vertical. The potentiallv warm air would be 

 at the top instead of at the bottom, and the effect of a 

 completed process of convection would be that pres- 

 sure would rise within the area of operations. But 

 if the risen air were delivered into a rapidly moving 

 current at the top, the air which it had •'evicted" 

 from the environment on its way would be lost to 

 the column, and when the process was completed the 

 air would close in from the top. the bottom, and the 

 sides. If there were any relative motion to begin 

 with— and there is alwavs some — closing in from the 

 sides must develoo cyclonic circulation with a cold 

 core. Closing in from the bottom with air colder 

 and drier than that which began the convection would 

 stop on account of dvnamiral cooling, and closing in 

 froip the top means the settling down of the air of 

 the stratosphere and a consequent low tropopause 

 with a .column of air above it warmer than its 

 environment. 



These conditions describe what the late Lord Rav- 

 leigh postulated for superposing a vortex on a current 

 with relative velority of its parts. They also agree 

 with what Mr. Dines describes as the results of his 

 examination of actual cyclonic conditions in Eng- 

 land. And this view of the procedure is borne out 

 by the examination of tropical cyclones. We can 

 form legitimate inferences from the pressure 

 records of these visitations because the normal 

 conditions of the localities where thev occur are 

 extremelv regular. We can see bv an inspection of 

 the graph of nressure that the region covered bv a 

 cyclone has simolv lost a certain part of the air which 

 it normally possesses. In one example I estimated the 

 loss as equivalent to 40,000 cubic km. at sea-level. 

 Beyond all doubt or question air had gone ; it was 

 not piled up in antirvrlones fore and aft. as we used 

 to think the convected air of our cvclones must be ; 

 it was gone clean awav. I suspect that it travelled 

 away in some unt>er current until slowed down over 

 the tropical antipvlone of some ocean. The story 

 will not be complete imtil that surmise is verified or 

 the correct account substi»iite<I. Hence, for the time 

 being, I am as curious about the life-historv of rnn- 

 vective air-currents as I was twenty years ago about 

 that of surface air-rurrents. 



In anv case, it seems to me certain that, becau.se 

 it carries away part of the air which it meets on its 

 path, convection, wherever it occurs, must entail 

 convergence, and therefore, exiepi at the equator, it 

 must give rise to a cyclonic circulation which may be 

 transient or, if rirrumsfances arc favourable, per- 

 manent. The function of the stratosphere seems to 

 be not constructive, but conservative and repisfrative. 

 It protects the energy from bein" dissipated bv 

 "filling up," because the descent of its isothermal 

 air is arrested bv the .idi.Tbntir rise of temperature. 



NO. 2666, VOL. 106] 



That is, indeed, the common function of all "decks" 

 or lids in the atmosphere, of which the stratosphere 

 is the chief. At the same time, for an observer the 

 stratosphere registers the locality of low pressure 

 by the towness of the tropopause and the relative 

 warmth of the air column above it. It seems to be a 

 law for the general circulation and for local circula- 

 tions that as pressure diminishes in the troposphere 

 the tropopause is lowered and the temperature of the 

 columns above it rises. 



Consequently, my view at the present time is that 

 the energy of a cyclone is due originally to convection 

 in a region with a suitable law of variation of velo- 

 city with height; it is guarded at the top by the 

 isothermal condition of the stratosphere, and on the 

 sides by the balance of pressure and rotation. It is 

 open to slow attack at the bottom on account of the 

 friction of its winds with the surface, and unless its 

 energy can be maintained by additional convection 

 It must perish. I do not think that a travelling 

 cyclon" carries its supply of rain for long distances; 

 it probably manufactures it out of the material in 

 the lowest levels which it has to pass over. But it 

 uses the energy so .supplied first to form a secondary, 

 and afterwards to absorb it or to be absorbed by it. 



Napier Shaw. 



Imperial College of Science and 

 Technology, S.W.7. 



It is a well-known hydrodynamical result that, in 

 the absence of any external stabilising influence, any 

 surface of discontinuity of velocity in a fluid must be 

 unstable. The effect of this instability is seen in the 

 eddies produced in a millpond, at the margin of the 

 entering stream. \ sufficiently rapid shearmg, with- 

 out actual discontinuity, will produce the same effect. 

 Most atmospheric eddies are developed in this way. 

 In the case of differences of velocity between different 

 masses of air at the .same level, gravity is not directly 

 available to damp any eddies that may be produced, 

 and hence it does not seem likely to be difficult to 

 account for eddies with their axes vertical. 



Thus the origin of cvclones may well be explained 

 on the lines suggested in Mr. W. H. Dines's letter in 

 Nature of November 18. It is rather more difficult to 

 see what determines the size and intensity to which 

 they grow. Ground friction must play its part ; also, 

 where the warm stream on the soiith side bulges 

 northward, it must do so to some extent over the 

 too of the cold air already there, and this arrangement 

 makes for stability, and when sufficiently developed 

 must prevent the further growth of the disturbance. 



The speed of translation of the cvclone on this 

 theory should be the mean of the velocities of the 

 two currents, which is usually about correct. The 

 geostroohic condition must also hold approximately, 

 otherwise the disturb.nnce would spread out with 

 nearly the velocity of sound and disappear. What is 

 not easy to see, however, is whv the isobars tend to 

 become more or less circular instead of wavy. 



Harold Jeffreys. 



Meteorological Offirc^, .South Kensington. 



I SHOULD like to express my agreement with Mr. W. H. 

 Dines's view (Naturp., November 18, p. 375) regarding 

 the origin of the initial difference of pressure which 

 leads to the development, under the influence of the 

 earth's rotation, of cyclonic circulation, and to state 

 that I have often suggeste<i that this initial disturb- 

 ance may have a mechanical origin (see Quart. Journ. 

 Rov. Meteor. Soc.. vol. xliii., iQt?, p. 27). At the 

 same time it st(iii». that one cannot, on many 

 grounds, ignon th f temperature contrnstn «'« 



