elphinium
Ajacis_ (the latter only known thirty miles off in cornfields in
Cambridgeshire), with _Atriplex patula_ and _A. deltoidea_. Gradually the
native sand plants--Carices, Grasses, _Galium verum_, &c., established
themselves, and year by year covered more ground till the new introductions
almost completely disappeared. The same phenomenon was observed in
Cambridgeshire between Chesterton and Newmarket, where, the soil being
different, _Stellaria media_ and other annuals appeared in large patches;
but these soon gave way to a permanent vegetation of grasses, composites,
&c., so that in the third year no _Stellaria_ was to be seen.

5. Mr. T. Kirk (writing in 1878) states that--"in Auckland, where a dense
sward of grass is soon formed, single specimens of the European milk
Thistle (_Carduus marianus_) have been known for the past fifteen years;
but although they seeded freely, the seeds had no opportunity of
germinating, so that the thistle did not spread. A remarkable exception to
this rule occurred during the formation of the Onehunga railway, where a
few seeds fell on disturbed soil, grew up and flowered. The railway works
being suspended, the plant increased rapidly, and spread wherever it could
find disturbed soil."

Again:--"The fiddle-dock (_Rumex pulcher_) occurs in great abundance on the
formation of new streets, &c., but soon becomes comparatively rare. It
seems probable that it was one of the earliest plants naturalised here, but
that it partially died out, its buried seeds retaining their vitality."

_Medicago sativa_ and _Apium graveolens_, are also noted as escapes from
cultivation which maintain themselves for a time but soon die out.
(_Transactions of the New Zealand Institute_, Vol. X. p. 367.)

The preceding examples of the _temporary_ establishment of plants on newly
exposed soil, often at considerable distances from the localities they
usually inhabit, might, no doubt, by further inquiry be greatly multiplied;
but, unfortunately, the phenomenon has received little attention, and is
not even referred to in the elaborate work of De Candolle (_Geographie
Botanique Raisonnee_) in which almost every other aspect of the dispersion
and distribution of plants is fully discussed. Enough has been advanced,
however, to show that it is of constant occurrence, and from the point of
view here advocated it becomes of great importance in explaining the almost
world-wide distribution of many common plants of th