On the Surface Temperature of North Sea and North Atlantic. 9 



(3) Interpolated positions (in Long. West), along the parallel of 41" N., wliicli 

 are intersected by the respective Isothermal Lines (Fahr. and Cent.). 



It is of very considerable importance that l>y this method we 

 can obtain the results which we are seeking in any particular 

 scale {e.g. Centigrade) without converting our whole data from the 

 one scale to the other ; not only is much time saved, but we avoid 

 the many opportttnities for inaccuracy which beset us in converting 

 a long series of round numbers from one scale into another : 

 for experience shows that we are apt to get into difficulties in such 

 a process, whether we convert the round numbers of one scale into 

 round numbers in the other, or whether (with an appearance of 

 greater, but only spurious, accuracy) we convert the round numbers 

 of the one scale into precise decimal eqttivalents in the other. 



After we have followed out the process here described for every 

 degree of latitude of which we have information, we may then at 

 once lay down upon our chart, not the actual observed temperatures, 

 but simply the ascertained positions where the isotherms cross the 

 parallels of latitude ; and these points are then immediately con- 

 nected up, so as to give us the complete isothermal lines. It will 

 usually be well worth while to repeat the whole operation in the 

 other direction, taking now the serial temper3,tures along each 

 meridian. And the two charts which we thus obtain, one from the 

 fluctuations in latitude and the other from the same fluctuations 

 studied in longitude, will materially help to check and correct one 

 another. 



But we have yet another check which we may with advantage 

 apply, in the case of our ^Monthly Mean Temperatures. Our twelve 

 monthly charts (which we may suppose that we have now drawn) 

 show twelve phases of one common phenomenon, throughout which 

 everywhere a principle of continuity runs. We shall expect the 

 temperature distribution in each month to be in harmony with that 

 of the Y)receding and with that of the succeeding month ; and if such 

 harmony is not apparent in our charts, the lack of it may perhaps 

 make us aware of some new and unexpected phenomenon, but is on 

 the whole much more likely to reveal to us some error or imj)erfection 

 in the systems of isotherms which we have drawn. Accordingly, we 

 proceed to construct a series of isopleth diagrams, such as those in 

 Figs. 25 to oO, each diagram showing the temperatures along one 



