MAGNETIC VARIATION. 147 



the interior of the small Sandal-wood Island, 91 in a direct 

 east and west direction from about 120 30' to 93 30' E 

 long., as had been accurately demonstrated sixteen years 

 before by Barlow. From the last named meridian it ascends 

 towards the north-west in 9 30' S. lat., judging by the posi- 

 tion in which Elliot followed the curve of 1 east variation 

 to Madras. We are not able here to decide definitely whether, 

 crossing the equator in about the meridian of Ceylon, it- 

 enters the continent of Asia between the Gulf of Cambay and 

 Guzurat, or further west in the Bay of Muscat, 92 and whether, 

 therefore, it is identical 93 with the curve of no variation, 

 which appears to advance southward from the basin of 

 the Caspian Sea ; or whether, as Erman maintains, it may 

 not curve to the eastward, and rising between Borneo and 

 Malacca, reach the Sea of Japan, 94 and penetrate into Eastern 



91 Elliot, in the Phil. Transact, for 1851, pt. i, p. 331, pi. xiii. The 

 long and narrow small island from which we obtain the sandalwood 

 (tschendana, Malay and Java, tschandana, Sanscrit, fsandel, Arab). 



92 According to Barlow, and the chart of Lines of Magnetic Declina- 

 tions computed according to the theory of Mr. Gauss, in the Report of the 

 Committee for the Antarctic Expedition, 1840. According to Barlow the 

 line of no variation proceeding from Australia enters the Asiatic Con- 



'tinent at the Bay of Cambay, but turns immediately to the north-east, 

 across Thibet and China, near Thaiwan (Formosa), from whence it 

 enters the Sea of Japan. According to Gatiss, the Australian line 

 ascends merely through Persia, past Nishnei-Nowgorod to Lapland. 

 This great geometrician regards the Japan and Philippine line of no 

 variation, as well as the closed oval group in Eastern Asia, as entirely 

 independent of the line belonging to Australia, the Indian Ocean, 

 Western Asia, and Lapland. 



93 I have already elsewhere spoken of this identity, which is based upon 

 my own declination-observations in the Caspian Sea, at Uralsk on the 

 Jaik, and in the Steppe of Elton Lake (Asie Centrale, t. iii, pp. 458 461). 



94 Adolf Erman's Map of the Magnetic Declination, 1827 1830. 

 Elliot's chart shows, however, most distinctly that the Australian curve 

 of no variation does not intersect Java, but runs parallel with, and at a 

 distance of 1 30' latitude from the southern coast. Since, according 

 to Erman, although not according to Gauss, the Australian line of no 

 variation between Malacca and Borneo enters the Continent through 

 the Japanese Sea, proceeding to the closed oval group of Eastern Asia, 

 on the northern coast of the Sea of Ochotsk (59 30' N. lat.), and again 

 descends through Malacca, the ascending line can only be 11 distant 

 from the descending curve ; and according to this graphical representa- 

 tion, the Western Asiatic line of no variation (from the Caspian Sea to 

 llussian Lapland) would be the shortest and most direct prolongation 

 of the part descending from north to south. 



L 2 



