From Charles Lyell¹   8 September 1860

Your argument from absence of mammalia in islands excepting those which we know to have been joined to mainland in Pliocene times is a favourite one of mine when I wish to prove that y^r bond of descent is required for the coming in of new species.² The older the island, the stronger yr case against special creation without a closely allied antetype as Wallace would say.³

The land shells of Madeira and the Canaries are so endemic as to show that none were united with each other or Africa. I believe all were separate from Upper Miocene times; if not, small African species of quadrupeds would have been retained.

But the antiquity of islands void of mammals checks the hypothesis of the easy adaptability of one species of a genus to new conditions. On a new volcanic island before all the best places are seized upon, seals or walrus’s, manatees, dolphins & other cetacea, when pressed hard for food w^d go up the rivers or if amphibious devour the eggs or young of land-birds. Bats, though not primates, but here Lissencephala ought to get the better in some stations of marine birds if higher classes tend to become dominant.⁴

The long reign of the Dinornis family in New Zealand implies that for ages there had been room for mammalia if bats & rodentia & aquatic placentals being at hand had been convertible even into allied structures fitted for land habits. And why bats & rodents should have peopled all Australia without having been developed into something higher in the placental line seeing that the representatives of the latter were imported from Europe, can run wild there, is strange, for why sh^d it take more time for a bat to work up into a lemur than for a Myrmecobius to improve into a Thylacinus.⁵

As to the land shells of Madeira & Porto Santo 95 per cent of them have remained absolutely unchanged for a period sufficiently long to allow the form & size of the islands to be altered materially by the waves of the Atlantic, long enough for some species which once very common to become almost extinct—in a few quite lost

The only speculation I have been able to make is this that by far the majority $\frac{9}{10}$ ^ths perhaps are immutable, & when changes come, must die rather than yield. The hereditary power has got too strong a set, after a million of years, in one way    It cannot go in any other. But there are some eminently metamorphic species in the Madeiras & these are producing the allied species & sub-species of Lowe in each island.⁶ & after a long period there might be found, in spite of extinctions, just as many true species of Helices in the Archipelago as now, supposing Man would let it alone—   So I suspect it has always been— Most species immutable & true to the death, as I maintained in the Principles that all were, but a few of the whole plastic & becoming the centers of new generic & ultimately higher groups as you maintain. So it seems to me to hold good in the tertiary series    Some shells once abundant in the Mediterranean formation & in the Crag, still survive unaltered though excessively persecuted & rare.—

Y^r antagonists are driven to great straits when they try to set up any machinery other than that of normal generation— The greater we can show the role of extinction, the worse is their case because the reactivating power ought to keep pace, if as Bronn pretends, the creation is always growing richer & more varied,⁷ which however I doubt since Eocene days.

According to the progressive theory, why sh^d. there be a living platypus or Ornithorhynchus, for we have nothing synthetical or elementary in the Coal Strata, & if they sh^d. be found there, how have they escaped being altered, improved and specialized in 30 periods.

Ammonites in uppermost part of the Maestricht, of a Cretaceous type, if not species of Texas are against the doctrine of wide gaps. Between Maestricht & Thanet sands.

What you say of the power of preoccupancy is good⁸— so, as to atolls, but if they subsided very slowly, the absence of volant & amphibious forms of reptile & mammifer in such a region proves rate of transmutation slow, even as compared to revolutions in physical geography.