48 
Choice Ferns for Amateurs. 
as a Fernery in a natural ravine with tortuous paths. 
These, however, are too extensive to be thought of 
except by the wealthy, still they are recorded as 
object-lessons. Yery important in the case of an 
excavated hardy Fernery (Fig. 31), even on a small 
scale, is the provision for getting rid of the surplus 
moisture — it must be carried away by drains. 
Even in making a Fernery on the level, the 
idea must be to provide good-sized pieces of rock, 
between the chinks and pathways of which certain 
suitable species may be introduced. No one would 
think of planting a Fernery on a dead-level for 
giving the best effect. That many have to confine 
their attention to a border or a bed is perfectly true ; 
but neither the one nor the other is an ideal 
place for Ferns. The reason of this is that the 
ordinary occupants thereof require altogether 
different conditions from those suited to Fern life. 
Sometimes it is possible to utilise very effectively 
old tree-stumps in combination with the stone ; but 
after all, it is doubtful Avhether decaying wood, with 
its great liability to be attacked with many insidious 
fungoid diseases, offers even a good environment 
for Ferns. Therefore, while on the one hand we 
would scarcely advocate the removal of such stumps 
did they exist on the other we would prefer relying, 
had we to form an entirely fresh rock garden, upon 
stone alone. Here it may be as well to state that it 
is very necessary to make the pieces of rock used in 
the construction of a Fernery firm in the soil, other- 
wise a heavy storm might prove disastrous. 
Our native Ferns alone provide sufficient variety 
^o satisfy the tastes of even the most exacting; but 
if greater variety be required then some of the exotic 
sr)ecies from Japan and North j^merica may be em- 
ployed, as their hardiness is undoubted. A mistake 
that is often made by amateurs is to plant without 
any idea as to the future development of the occu- 
pants. To take British Ferns as a guide, it is most 
certainly desirable to allow sufficient spare for them 
at the outset. Another point is to plani the back- 
ground in mass^yS instead of singly. If, for instance, 
