NO. 



4 TEMPERATURE VARIATIONS IN THE NORTH ATLANTIC 29I 



those of the northern and eastern Indian Ocean, and are remarkably 

 smaller. They may be a mixture between the latter fluctuations 

 and the inverse fluctuations occurring in the high-pressure region 

 of the South Atlantic to the west. 



Curve 16, figure io8, giving the twelve-month means of the pres- 

 sure at Stykkisholm, Iceland, shows that there is much direct simi- 

 larity between the barometric fluctuations of the Icelandic pressure 

 minimum and those of the Indo-Malayan low-pressure region, and 

 the Australian high-pressure region, situated very nearly anti- 

 podically. But the fluctuation in the Icelandic region is very much 

 greater. This may probably be due to the fact that the area of the 

 Icelandic pressure minimum is very small as compared with that 

 of the Indo-Malayan, Australian region. 



If we go only short distances outside the area of the Icelandic 

 minimum, e. g., to Aberdeen (Scotland), or to Norway, or to the 

 west coast of Greenland the barometric fluctuations differ very 

 much from those of Stykkisholm, and also from those of Ponta 

 Delgada ; the reason being that these regions are outside the cen- 

 ters of action and their fluctuations belong to a mixed type. 



The pressure curves VI, VII, and VIII (fig. 107) for Santiago 

 in Chile, Goya and Cordoba in Argentina, show great similarity to 

 the inverted curves IV and V for Batavia and Bombay. This is 

 in perfect accordance with what the two Lockyers liave found, as 

 we have mentioned in chapter X. 



The three South American stations are in the high-pressure belt 

 of the southern hemisphere between the maxima of the South Atlan- 

 tic and the South Pacific. The curve of Santiago shows the most 

 typical agreement with those of Batavia and Bombay, possibly 

 because it is nearer to the center of action, the annual pressure 

 maximum of the South Pacific, than the two" Argentina stations are 

 to that of the South Atlantic. 



The two types of curves {e. g., fig. 107, curve VI, and curves I-III ; 

 fig. 108, curve 16 and curves 2-15 show on the whole much similarity 

 to each other, but also dissimilarities, e. g., in the years 1877-78 

 (marked E), 1884-85, 1892-93, and partly 1894-95, though, e. g., 

 in 1894-96 the curve of Stykkisholm agrees remarkably well with 

 those of southern Australia (cf. fig. 108, curves 14, 15, 16). 



We have also made an analysis of the yearly barometric changes 

 in Siberia where, however, the conditions differ greatly during the 

 year, there being a high barometric maximum in southern central 

 Siberia and in Mongolia, in the winter, but a minimum in the 



