On March 3, 2003 Glen Marshall and I measured the thickness of a piece of aluminized mylar. The piece we measured came from a roll with the same specs as we used for the cathode foils (nominally 6.25um). The measurement is documented in TWIST Log Book #4 (start date 2002-05-27).
With Steve's help we cut a square piece of the foil about 60x60cm^2, no tension, small wrinkles. The weight of the piece was found to be (3.006+-0.002)g, that gives 8.35x10^-4g/cm^2. We estimated the error on the measurement to be about 0.3% which is mainly due to the uncertainty in the shape and size of the piece. (See the log book.)
We also measured the foil thickness directly with a micrometer. One layer is 0.00029'', four layers are 0.0011'' (translating into 0.00028'' per layer). The instrument accuracy is 0.000005'', so this measurement is to no better than 5%. Also a measurement of g/cm^2 is more relevant for GEANT, so we will ignore the micrometer result.
From the surface mass density assuming the specific density of mylar of 1.39g/cm^3 and ignoring the aluminum we get the effective thickness of 6.007x10^-4cm.
Most (?) of our cathode foils are stretched to 0.6% (a linear, not a surface factor). So the effective thickness of the foil in the detector is 6.007/(1.+0.006)^2=(5.936+-0.018)x10^-4cm, where the 0.3% error has been applied.
The thickness listed in dt_geo.00032 is 5.872x10^-4cm, which is 3.5 sigma below the new measurement.