SCHISTOCERCA, SPHINGONOTUS AND HALMENUS 415 



RELATIONSHIPS AND EVOLUTION OF THE RACES. 



The characters of the species and their various races show little 

 with regard to their interrelationships. If we leave out of consider- 

 ation the Duncan specimens, there is no difficulty in referring the 

 numerous other varieties to two well-separated species as has been 

 done. Since, however, the Duncan specimens completely bridge over 

 the wide gap between these two otherwise apparent species, we have 

 a less simple problem to deal with. 



Concerning the Duncan form we may make three suppositions as 

 follows : ( i) It may be an ancestral form, from one of whose extremes 

 has been developed 6". nielanocera and from the other S. literosa ; 

 (3) it may be a variable race that on the one hand has varied toward 

 S. melanocera and on the other toward S. literosa; (3) it may be a 

 hybrid race formed by the intermingling of representatives of the other 

 two species from the other islands. 



There are several objections to ( i ) . In the first place Duncan 

 appears least likely of all the islands to have received the original 

 Schistocerca population. It is situated near the center of the archi- 

 pelago and is not more than three miles in its longest diameter. 

 Structurally it consists of a circular extinct volcano, having an ex- 

 centric crater with a floor about a third of a mile in diameter and 475 

 feet above sea level. The rim of the crater is still perfect, rising in 

 most places to a height of about 350 feet above the floor. At one 

 point it is produced into a sharp peak having a height of about 1,000 

 feet. The island is dry and barren. The whole outer slope is covered 

 by a thick growth of low bushes, but soil is present only on the floor 

 of the crater. Animal life is almost entirely confined to the interior of 

 the crater. It is on this small area, affording absolutely uniform en- 

 vironmental conditions, that the most variable form of ScJiistoccrca in 

 the whole archipelago is found. It is highly improbable that an 

 island of this sort could have been fit for habitation earlier than such 

 fertile and much disintegrated islands as Chatham and Charles. More- 

 over, the winds and currents are invariably from the southeast. 

 Hence, Duncan is in a poor location to send migrants to the islands 

 south and east of it. Furthermore, it is impossible to imagine any 

 reason why variations representing the literosa form should have gone 

 to Tower, Chatham and Hood and not to any of the neighboring 

 islands. On the other hand, Duncan gets but little rain, and tlie ap- 

 pearance of an island cannot be taken as a definite indication of its age. 

 The presence of a very strongly marked race of land tortoise on Dun- 

 can might be taken as evidence of a considerable age for the island. 



