450 Mr. Wallace on the Geographical Distribution of Birds. 



which must certainly be a mistake. You give 12,000,000 

 square miles. Now Balbi gives only 8,500,000 for all Africa and 

 Madagascar ; and, if we estimate the part of Arabia taken from 

 Asia as equal to the part of Africa given to Europe, this will be 

 the true area. But I would suggest that such an anomalous 

 tract as the Sahara coming between two regions should be given 

 to neither; it should in fact be considered as a Sea. It is cer- 

 tainly quite as unproductive of animal life as the sea, perhaps 

 more so ; and it gives quite an erroneous idea of the productive- 

 ness of Tropical Africa to add this immense desert to it. I take, 

 therefore, Africa south of the Sahara, and after reaching the Nile 

 as far north as the 1st Region, following a diagonal across Arabia 

 from Mount Sinai to the eastern extremity. A careful measure- 

 ment gives me the area of this with Madagascar as 6,500,000 

 square miles. 



3. Indian Region. Of this we have already defined the north 

 limit, and I would add a tract of Arabia on the western shores 

 of the Persian Gulf. Its south-eastern limits I draw between 

 the islands of Bali and Lombok, and between Celebes and 

 Borneo, and the Moluccas and the Philippines. Barbets reach 

 Bali, but not Lombok; Cacatua and Troindorhynchus reach 

 Lombok, but not Bali : this 1 think settles that point. Ca- 

 catua, Trichoglossus, and Scythrops in Celebes, and not in Borneo, 

 settle the other. No doubt many Indian forms reach Celebes ; 

 but we must remember the proximity, and in the course of ages 

 the only wonder is thei'e has not been more intercommunication. 

 A careful estimate of these islands, with Formosa, &c. added to the 

 continental portions, gives an area of only 3, 100,000 square miles. 



4. Australian Region. You have rather over-estimated this : 

 with the Moluccas, &c., North Guinea, New Zealand, and the 

 Pacific Islands (except the Sandwich, which I think should go 

 with America), I cannot make an area of more than 2,600,000 

 square miles. In the island of Batchian I have found Podargtis, 

 Coriphilus, and Paradisea, which shows that the several sub- 

 divisions of this region are very closely connected. 



5 and 6. North and South American Regions. I put the 

 limit between these at 22° N.; on the coast it may be further 

 north, on the table-land further south, but this will be near the 



