240 Proceedings of Royal Society of Edinburgh. [sess. 
separately audible ; but they still exist in a sub-sensible form, and 
constitute no negligible fraction of the duration of the phonogram. 
Consider that the whole duration of the s may be only sec., and 
compare that with any possible rapidity of the transitional motions 
which create the glide : the glide must be, in this case at least, 
half as long as the consonant. How is it, then, that the glide is not 
separately heard? This arises chiefly from the persistence of 
auditory sensations. The sensation of the vowel does not sub- 
jectively cease at the point where the vowel vibrations terminate 
in the phonogram. It persists during the whole duration of the 
glide, and practically blots it out by its more powerful timbre. It 
follows, however, from the principle that the duration of the glide 
is the duration of the longest movement involved, that glides are 
not normally of at all equal length. The combination bl in blue, 
already instanced, is practically glideless, because the l articulation 
can be, and is, framed inside the b articulation, and simultaneously 
with it. In combinations of h + vowel, there is a real glide, 
because the larynx must have time to shut before the h can be 
transformed into a toned vowel ; and there is inevitably a certain 
portion of this time during which the h has been spoiled by the 
narrowing of the glottis, though the chords are not yet vibrating, 
and the toned vowel is, therefore, not yet begun. But this period 
must be exceedingly short • and in speaking of such combinations 
in a former passage we have neglected it (see also infra). But 
sometimes the transitional movement is much more cumbrous than 
these, and takes much longer time. At times, indeed, it is im- 
possible, by any attainable rapidity, to prevent the glide being 
audible to a quick ear. Take the English word, eel. During the 
ce the tongue is presented convex to the hard palate, with a passage 
of some 50 sq. mm. section between them. For l this must be 
entirely changed. The tongue-tip must be withdrawn to the 
alveolars ; the passage must be shut up, and new passages opened 
at each side ; the convex curvature of the tongue-blade must be 
exchanged for one slightly concave. Hence the sound of the ee 
and the sound of the l are never contiguous, and the dull interven- 
ing sound is always long enough to be separately apprehended by 
a quick ear. In French fit this is not quite the case. The French 
l is “ dorsal,” not “ coronal,” i.e., it is articulated with the blade, 
