Update from my post in 2020. When I put my instrument's caps on a bridge, C91 and C94 (made by Erie) read like polyester, while C92 and C93 (Sprague 160P "DiFilm" paper/plastic) have appreciable Dielectric Absorption, which shows up as increasing C as F is reduced. When I replaced C92 and C93 with modern plastic-film caps, I found that I overshot the mark - where the 130 previously read high at the low end of the scale, now it reads low at the low end of the scale. Just a bit, enough to annoy me. It was not due to polypropylene, I stuck in a polyester and there was no perceptible change. I decided to make a first-order simulation of olde-tyme paper DA by shunting the cap with a series RC. Using a substitution box I groped around for a sweet spot. I found one for C92 (.0047uF, 30pF range), 470pF + 470K. YOUR MILEAGE WILL VARY. The time constant t is 0.22ms which in a filter transitions at 1/(2 * pi * t) = 0.72kHz or about mid-scale frequency.
Low/mid/high readings with 11.3pF Cx.
Before: 11.5/11.8/12.5
After: 11.5/11.0/11.5
I could not get enough improvement on C93 or C94 to bother, but I was able to help C91 (.0015uF, 100pF range) with 220pF + 150K.
As you can see by the pictures at TekWiki, when Tek started making the 130 in 1959, they built it with paper-oil "Bumblebee" caps which had more DA than polyester even when new. As time passed, the Purchased Item Source List (PISL) changed the various part numbers from paper to paper/plastic and then full plastic, but the meter scale was still tooled for paper.
