OF THE AMAZON VALLEY. 
531 
inquired into*. I found no clear instance on the banks of the Amazons of two forms 
of M. Polymnia having this relationship; but, to judge from the collections received, it 
exists between M. Lysimnia and M. Nescea, in the middle parts of the Brazilian coast- 
country, at Bahia. 
Local var. Mechanitis Egaensis (PI. LVI. fig. 7 a). 
Expanse, 6 2" 2"'*-2" 7'", 2 2" 4/"-3". tf . Fore wing similar in ^hape to that of M. 
Polymnia : above, dark orange-tawny or reddish brown ; the basal part of the costa and 
a broad stripe along the hind margin, extending three-fourths the length of the wing, and 
touching the edge, black ; there are, besides, six black spots, namely, a large quadrate one 
about the middle of the cell, a faintly marked one near the base, two (sometimes united) 
over the end of the cell, and two between the 1st and 2nd median branches. Across the 
wing, a little beyond the middle, is a very irregular bright-yellow belt, which begins on 
the costa and terminates about the middle of the outer margin: this band generally 
includes a quadrate spot lying within the end of the cell; its outer margin is deeply 
bisinuated, and broadly edged with black: in the middle of the brownish apical part of 
the wing, near the costa, is a large ill-defined yellowish spot. Beneath, the same, except 
that there is on the outer margin, near the apex, a scries of five whitish spots, in some 
examples reduced in size and number, in others altogether wanting. 
Kind wing short and quadrangular, the angularity most pronounced in small examples : 
above, same colour as the fore wing; there is a broad, black, zigzag stripe along the disk, 
found 
behind the cell, and generally a series of marginal lunules of the same colour. The 
marginal lunules are wanting in some examples ; in others they are large, and blended 
partly with the black discal stripe. Beneath, the same, the margin spotless ; there is a 
yellow spot at the base of the costa. 
? . Similar to the <$ in colours : the hind wing is not angular, but elliptical in shape, 
and has generally a trace of the marginal white spots on the under surface. Individuals 
occur much darker in colour, with the black marks spreading partly over the surface. 
Extremely abundant all the year round at Ega, in thinned parts of the forest ; but 
in no other part of the Amazon region. It occurred in company with the typical M. 
Polymnia and numerous individuals exhibiting all the gradations between the two 
extreme forms. I add descriptions of some of these intermediate varieties. 
* I have given a case ill the first paper published of thi- series, "On the Insect Fauna of the Amazon Valley," in 
• Transactions of the Entomological Society of London,' n. s. vol. v. p. 3:. i, Papilio Vertumnns and P. Hierocles. It 
seems to me that, as only those local varieties become species which maintain themselves distinct when brought bj 
redistribution in contact with their sister-forms, natural selection comes here into play. It is an advantage to a 
form to have a sphere of life different from its allies : when two sister forms keep themselves distinct in a locality, it 
is a sign they have acquired sufficient difference to fill two separate spheres ; if they paired together, they would soon 
become one again. Nature may be said to place a premium on diversity ; for she thus destroys the incompletely formed 
race, and preserves the completely formed one. 
The case of Mechanitis Polymnia differs from that of Leptalis Theonoe, in exhibiting the production, generally, of 
only one local form in a district, instead of many. As far as my observations go, this seems to have been the most 
frequent course in nature. More than one new race would with difficulty be formed in a limited area, where the 
individuals live in close neighbourhood, except in such cases as our Leptalis, where rigid destruction of intermediate 
forms is going on, thus restricting the choice of mates to the surviving forms, or in such genera as Mom* where 
there is no doubt the insects carefully select their exact counterparts in pairing. 
