ifice at the side, a sort of round dog-hole, just
large enough for the insect to pass through. When the cell is thus
marked out and almost wholly closed, the Osmia attends to the storing of
her provisions and the laying of her eggs. Steadying herself against the
margin of the hole at one time with her fore-legs and at another with
her hind-legs, she is able to empty her crop and to brush her abdomen;
by pressing against it, she obtains a foothold for her little efforts
in these various operations. When the tube was narrow, the outer wall
supplied this foothold and the earthen partition was postponed until the
heap of provisions was completed and surmounted by the egg; but in
the present case the passage is too wide and would leave the insect
floundering helplessly in space, so the partition with its serving-hatch
takes precedence of the victuals. This method is a little more expensive
than the other, first in materials, because of the diameter of the reed,
and secondly in time, if only because of the dog-hole, a delicate piece
of mortar-work which is too soft at first and cannot be used until it
has dried and become harder. Therefore the Osmia, who is sparing of her
time and strength, accepts medium-sized reeds only when there are no
small ones available.

The large tubes she will use only in grave emergencies and I am unable
to state exactly what these exceptional circumstances are. Perhaps she
decides to make use of those roomy dwellings when the eggs have to be
laid at once and there is no other shelter in the neighbourhood. While
my cylinder-hives gave me plenty of well-filled reeds of the first and
second class, they provided me with but half-a-dozen at most of the
third, notwithstanding my precaution to furnish the apparatus with a
varied assortment.

The Osmia's repugnance to big cylinders is quite justified. The work in
fact is longer and more costly when the tubes are wide. An inspection of
a nest constructed under these conditions is enough to convince us. It
now consists not of a string of chambers obtained by simple transverse
partitions, but of a confused heap of clumsy, many-sided compartments,
standing back to back, with a tendency to group themselves in storeys
without succeeding in doing so, because any regular arrangement would
mean that the ceilings possessed a span which it is not in the builder's
power to achieve. The edifice is not a geometrical masterpiece and it
is even less satisfactory from th