92 Mr W. N. Shaw and Mr F. M. Turner, [Feb. 28, 
result could be got with wave-lengths less than 15 mm. ‘This 
may be due to the fact that the area of the section of the flame 
becomes then too large in proportion to the wave-length of the | 
sound for the node to be identified, moreover the flame is then 
very sensitive, and there may be external disturbances which 
produce a continuous flaring if the flame is made sufficiently 
sensitive to respond to the action of the whistle, and which cannot 
be otherwise appreciated, or the particular flame used may be 
unsuitable for such very high notes. These suggestions require 
investigation before the fourth alternative is adopted, viz. that 
the action of the whistle is not continuous beyond the wave-length 
corresponding to about 3 mm. length of pipe. 
The notes emitted by three whistles (A, B, C) were tested 
by the method described by Lord Rayleigh. The pressure of 
the air blowing the whistle and the length of the whistle were 
both capable of being varied. The pressure of the gas supplying 
the sensitive flame could also be varied. 
Whether the flame flares or not in any given position depends 
not only on whether it is at a node or internode, but also on the 
state of the flame itself. If it is supplied by gas at too low a 
pressure it will not flare to any sounds, if at too high a pressure 
it flares spontaneously. If the pressure is raised from a low 
value upwards the flame first becomes sensitive to notes of wave- 
length about 15mm. For notes of long wave-length (up to about 
36 mm.) the flame has to be on the point of flaring, or even flaring 
slightly by itself.before it becomes sensitive. Not only does the 
flame vary in sensitiveness absolutely, but also as to the difference 
in its behaviour at nodes and loops, which greatly affects the 
experiments. In some cases it is possible by adjusting the 
pressure to get the flame to flare well in the loops and to keep 
steady or with only spasmodic flaring at the nodes. The range of 
the observations was limited by the fact that these favourable 
conditions could not be obtained at either extreme. In observing 
the two shortest wave-lengths (16 and 16-4mm.) the flame flared 
in all positions, but at the nodes the flaring was slightly but 
definitely less. With shorter wave-lengths no nodes were dis- 
tinguishable. Whether the whistle gave no definite note, or the 
fault lay with the flame is undecided. With notes of longer wave- 
length than 36 mm. the flame could be made to flare but not to 
shew distinct nodes. 
Experiments were also arranged to test (1) whether the formula, 
wave-length = 4x length of pipe, holds good; (2) whether the 
different whistles give the same note; (3) whether the note varies 
with the pressure of air blowing it. Instead of giving all the 
measurements one series is quoted as a sample, and a table of 
all the results placed after it. 
