When the Reynolds number is treated as an asymptotically large number
in a boundary-layer stability analysis, it is possible to identify Rey
nolds number scalings at which different terms in the governing equati
ons balance. These balances determine which physical mechanisms are op
erating under which circumstances and enable a systematic treatment of
the various parameter regimes to be carried out. Linear waves can gro
w by viscous mechanisms if the velocity profile is noninflexional and
both viscous and inviscid instabilities are present for inflexional pr
ofiles. When disturbances become nonlinear the resulting dynamics depe
nd upon the type of instability and, in particular, on whether the cri
tical layer lies within the viscous wall layer or is separate from it.
Therefore, the classification of a wave as viscous or inviscid is imp
ortant to the theory of transition. However, at finite Reynolds number
s the boundaries separating different types of instability become blur
red. Moreover, certain asymptotic theories are known to give poor quan
titative agreement with experiment while others remain untested by det
ailed experimental comparison. This paper is concerned with identifyin
g the domains where the different asymptotic theories are most relevan
t so as to facilitate their comparison with experiment. Numerical solu
tions of the Orr-Sommerfeld equation are presented that indicate that
in wind-tunnel experiments the instability driving transition is essen
tially viscous even for adverse pressure gradient boundary layers, and
that inviscid instability waves would be difficult to observe. For an
inviscid wave near the upper branch there could he a lower frequency
viscous wave at the same point in the how with amplitude 50 times larg
er in a typical practical situation. (C) Elsevier, Paris.