o more time in this darkness, which idle theorizing will
not dispel; let us return to facts, humble facts, the only ground that
does not give way under our feet. The Osmia respects her neighbour's
cocoon; and her scruples are so great that, after vainly trying to slip
between that cocoon and the wall, or else to open a lateral outlet, she
lets herself die in her cell rather than effect an egress by forcing
her way through the occupied cells. When the cocoon that blocks the way
contains a dead instead of a live grub, will the result be the same?

In my glass tubes, I let Osmia-cocoons containing a live grub alternate
with Osmia-cocoons in which the grub has been asphyxiated by the fumes
of sulphocarbonic acid. As usual, the storeys are separated by disks of
sorghum. The anchorites, when hatched, do not hesitate long. Once the
partition is pierced, they attack the dead cocoons, go right through
them, reducing the dead grub, now dry and shrivelled, to dust, and at
last emerge, after wrecking everything in their path. The dead cocoons,
therefore, are not spared; they are treated as would be any other
obstacle capable of attack by the mandibles. The Osmia looks upon them
as a mere barricade to be ruthlessly overturned. How is she apprised
that the cocoon, which has undergone no outward change, contains a dead
and not a live grub? It is certainly not by sight. Can it be by sense of
smell? I am always a little suspicious of that sense of smell of
which we do not know the seat and which we introduce on the slightest
provocation as a convenient explanation of that which may transcend our
explanatory powers.

My next test is made with a string of live cocoons. Of course, I cannot
take all these from the same species, for then the experiment would not
differ from the one which we have already witnessed; I take them
from two different species which leave their bramble-stem at separate
periods. Moreover, these cocoons must have nearly the same diameter to
allow of their being stacked in a tube without leaving an empty space
between them and the wall. The two species adopted are Solenius vagus,
which quits the bramble at the end of June, and Osmia detrita, which
comes a little earlier, in the first fortnight of the same month. I
therefore alternate Osmia-cocoons and Solenius-cocoons, with the
latter at the top of the series, either in glass tubes or between two
bramble-troughs joined into a cylinder.

The result of this promiscuity