I would forget trying to profile via your Spectroscan T. Your main
problem in trying to create an 'accurate' profile is anticipating the
colour temperature of the light source of the overhead projector (this
will vary with age and use), the reflective properties of the
projection screen and the influence of any ambient light in the
projection area. Since you are unlikely to have control over any of
these criteria I suggest your best option is to treat the transparency
as a reflective media and create your profile by placing a sheet of
white paper under the transparent material. You could try using a
couple of different 'white' media particularly one that looks 'yellow'
to anticipate the yellowness of a tungsten lamp.
If you have consistent colour (via your profile) in creating your OHP
transparencies then once they are projected the viewers' eyes will
adapt to the projection environment and 'see' the colours as closely as
possible to the intended.