Yol. 59.] THE EIGtTKE OF THE EARTH. 181 



Celebes, and another in Borneo, are bent round near their termi- 

 nation into approximate correspondence with the trend of the 

 circular arcs. 



A third example is afforded by the western shore of the southern 

 part of North America ; the centre in this case is situated in 

 Missouri, near lat. 37° N., long. 91° W. The arc is closely coinci- 

 dent with the coast of California and Mexico as far as Guatemala, 

 ending in a region which is the fruitful parent of world-shaking 

 earthquakes. (See A in fig. 2, p. 183.) 



Other instances, quite as excellent in their way, but on a smaller 

 scale, might be cited, though for our present purpose it will conduce 

 to brevity if we point to some less perfect correspondences. The 

 northern part of South America is one of these : a centre situated in 

 lat. 1° N. and long. 65° W. affords a circle which follows approxi- 

 mately the Andes of Peru, Ecuador, and their continuation into 

 New Granada : (see C in fig. 2). Similarly, a centre in lat. 18° N., 

 long. 3° W., gives a circle which more or less coincides with the 

 contour of Western North Africa : (see Af in fig. 1, p. 182). 



The great Alpine-Himalayan chain might at first sight be 

 regarded as far too irregular to submit of reduction to geometrical 

 form, but the case is not quite hopeless. The Himalayan chain 

 itself is remarkably circular, its centre lying in lat. 42° N.,, 

 long. 90° E. A circular arc having its centre at about lat. 69° N.,. 

 long. 89° E., sweeps through the Caucasus to join the Himalayas, the 

 intermediate chains between swaying symmetrically on each side of 

 it. A second great arc, with its centre close to Iceland in lat. 69° N. r 

 long. 18° W., falls upon the Atlas range, in complete coincidence, 

 and also upon the Crimean segment, at the termination of which it 

 meets and cuts the great Caucasian curve : (see I in fig. 1, p. 182). 

 A smaller circle from the same centre is nearly tangential to the 

 Northern Alps and the Carpathians. 



The centre in North Africa already mentioned affords a circle 

 which runs through the Betic Cordillera, and another on which the 

 Pyrenees lie ; the latter passes nearly symmetrically through the 

 Tyrrhenian Sea to traverse the site of Etna. This circle and 

 the three previous ones, which we may call the Atlas, Caucasian, 

 and Icelandic circles, enclose a lozenge-like space, within which 

 the folding of the Apennines, Alps, Carpathians, and Balkans 

 has taken place. Where the African overlaps the Atlas circle, 

 the sudden inflection of the Atlas guiding-line to that of the Betic 

 Cordilleras makes its appearance. In a similar way an Australian 

 circle, having its centre in the Australian bight, passes through the 

 length of New Guinea, and, where it cuts the grand East Indian 

 arc, produces the well-known inflection which terminates in Ceram : 

 (see Au in fig. 2, p. 183). 



The important question, for which we have now prepared the 

 way, will naturally be raised, whether it may not be possible to 

 bring the chief of the imaginary centres which have been indicated 

 into some general connexion. The two best-marked examples of a 



