descending streams--if the air below be warmed, we have
ascending streams as the initial cause of atmospheric flocculence. A
sound proceeding vertically does not cross the streams, nor impinge
upon the reflecting surfaces, as does a sound proceeding horizontally
across them. Aerial echoes, therefore, will not accompany the
vertical sound as they accompany the horizontal one. The experiment,
as I interpret it, is not opposed to the theory of these echoes which
I have ventured to enunciate. But, as I have indicated, not only to
see but to vary such an experiment is a necessary prelude to grasping
its full significance.

In a paper published in the 'Philosophical Transactions' for 1876,
Professor Osborne Reynolds refers to these echoes in the following
terms Without attempting to explain the reverberations and echoes
which have been observed, I will merely call attention to the fact
that in no case have I heard any attending the reports of the rockets,
[Footnote: These carried 12 oz. of gunpowder, which has been found by
Col. Fraser to require an iron case to produce an effective
explosion.] although they seem to have been invariable with the guns
and pistols. These facts suggest that the echoes are in some way
connected with the direction given to the sound. They are caused by
the voice, trumpets, and the syren, all of which give direction to the
sound; but I am not aware that they have ever been observed in the
case of a sound which has no direction of greatest intensity.' The
reference to the voice, and other references in his paper, cause me to
think that, in speaking of echoes, Professor Osborne Reynolds and
myself are dealing with different phenomena. Be that as it may, the
foregoing observations render it perfectly certain that the condition
as to direction here laid down is not necessary to the production of
the echoes.

There is not a feature connected with the aerial echoes which cannot
be brought out by experiments in the air of the laboratory. I have
recently made the following experiment: A rectangle, x Y (p. 331),
22 inches by 12, was crossed by twenty-three brass tubes (half the
number would suffice and only eleven are shown in the figure), each
having a slit along it from which gas can issue. In this way
twenty-three low flat flames were obtained. A sounding reed a fixed
in a short tube was placed at one end of the rectangle, and a
'sensitive flame,' [Footnote: Fully described in my 'Lectures