764 



NA TURE 



[December 9, 1922 



the changes of pressure and temperature for a point 

 near Munich and the changes of momentum at a 

 point between Munich and Hamburg are calculated 

 for the interval of six hours centred at 1910 May 2od. 

 7I1. G.M.T. That day was chosen for displaying the 

 method because a set of data for the surface and upper 

 air was available in the publications of the Geophysical 

 Institute of Leipzig issued by Professor V. Bjerknes. 

 Although not quite adequate for the purpose it is an 

 unusually full set. 



The calculation occupied " the best part of six 

 weeks " in a rest-billet in France. It included, however, 

 the preparation of the forms which are now issued in 

 blank for the use of others who may be attracted by 

 the prospect of submitting the course of Nature to 

 the process of numerical calculation. Every assistance 

 is given by the forms and by suggestions for improving 

 the accuracy, smoothing the data, and many other 

 technical points of manipulation. 



The trial specimen is not such a good example of 

 the art of forecasting that it tempts the reader forth- 

 with to become one of the great orchestra. The change 

 of pressure at the surface works out at 145 millibars 

 in six hours. Our barometers allow for a range of 

 100 millibars at most ; and, as a matter of observation, 

 the change in the region in question was less than 

 a millibar : the wildest guess, therefore, at the change 

 in this particular element would not have been wider 

 of the mark than the laborious calculation of six 

 weeks. Nor is that all. Many of the chapters end 

 in parenthetic expressions of regret or of suggestions 

 for improvement. There are also many supplementary 

 paragraphs which indicate that when the author comes 

 to make another edition, as he or some one else un- 

 doubtedly will, he will write somewhat differently. 

 And the reader will not be sorry, for in many ways 

 the book makes hard reading. It is full of mathe- 

 matical reasoning, a good deal of which is conducted 

 " by reference." The reader who wishes to follow 

 it must have a very handsome library and a few step- 

 ladders which Mr. Richardson does not provide. 



A reviewer with less than the ordinary sufferance 

 of his tribe might easily murmur : forecasting by 

 numerical process seems so arduous and so disappoint- 

 ing in the first attempts that the result is a sense of 

 warning rather than attraction. He might also wonder 

 for whom the author is writing, and regard the book 

 as a soliloquy on the scientific stage. The scenes are 

 too mathematical for the ordinary meteorologist to 

 take part in and too meteorological for the ordinary- 

 mathematician. But such complaint would be as 

 misleading as the computed forecast. On the road 

 to forecasting by numerical process nearly every 

 physical and dynamical process of the atmosphere 

 NO. 2771, VOL. I io] 



has to be scrutinised and evaluated ; the loss of view 

 into the future from the first summit is compensated 

 many times by the insight which one gets into the 

 working of Nature on the way. For example, the 

 author draws from the miss of his forecast the con- 

 clusion that the observations of velocity used are a 

 real source of error. Whether that conclusion is true 

 or not, its further consideration is of the greatest 

 importance in view of the multiplicity of observations 

 of winds in the upper air and of the difficulties which 

 their interpretation presents. 



The essential obstacle in the way of bringing the 

 facts of weather into mutual co-ordination by recognised 

 methods of dynamics and physics is that there are 

 so many of them, so many elements, so many variables, 

 so many causes of perturbation. Some meteorologists 

 look for a general solution of the problem in the dis- 

 covery of new physical laws, at present unthought of, 

 that will make things clear. Yet, even when we revel 

 in the proud consciousness of being familiar with all the 

 ultimate dynamical and physical laws to which the 

 atmosphere is subject, we may yet fail in an endeavour 

 to relate the conditions of the moment to those of 

 the past or to anticipate the future from the present 

 by lack of method in the arrangement of the 

 facts. 



When we look back at the triumphs of calculation 

 of the historic past we find always that the skilful 

 calculator has substituted an ideal, upon which it is 

 possible to operate, for the intractable reality. The 

 late Lord Rayleigh made the general position clear 

 in his first volume on " Sound," where he pointed 

 out that in order to study sound as vibration we 

 imagine the sounding body to be completely isolated, 

 though, if it were so, there would be no sound. Mr. 

 Richardson in his preface properly cites the Nautical 

 Almanac as an alluring example of forecasting by 

 numerical process. We are reminded of Plato's maxim, 

 " We shall pursue astronomy with the help of problems 

 just as we pursue geometry, but we shall let the heavenly 

 bodies alone if it be our desire to become really 

 acquainted with astronomy." Perhaps astronomers 

 have been disposed to press this maxim to the extreme, 

 yet we must admit that the Nautical Almanac owes 

 much to the ellipse in substitution for the actual 

 orbits of the heavenly bodies. It would perhaps be 

 difficult to imagine anything more unreal than the 

 latest ideal of the atom. 



Hence we might argue that the first step in meteoro- 

 logical theory should be to group the facts in such 

 a way as to replace the reality by a reasonable and 

 workable ideal. That view underlies the work of 

 Hildebrandsson and Teisserenc de Bort in " Les Bases 

 de la meteorologie dynamique," in which they 



