
Some months ago, I looked at the recommended weights for TT cars. First of all, the US NMRA RP 20.1 and European NEM 302 are wildly different – for a 40-foot TT car, NEM says (after the necessary unit conversions) 0.88 ounces, the NMRA says 2.25 oz! That's over a 2:1 difference, which btw holds for N and HO as well.
That caused me to look at the weight scaling across the modeling scales – here's the write-up I made for myself:
Weight – NMRA RP 20.1, in oz: ________________________________________ NEM 302: (+30% tol.) ???
N: __ 0.5 + 0.15 *L (in) ______________ 0.5 + 3*0.15 = 0.95 oz _________ 0,17 g/mm => 0.15 *3 = 0.45 oz
TT: _ 0.75 + 0.375*L - so 40' car => 0.75 + 4*0.375 = 2.25 ___________ 0,25 __________ 0.22 *4 = 0.88 !!
HO: _ 1.0 + 0.50 *L ___________________ 1 + 5.5*0.5 = 3.75 ____________ 0,40 __________ 0.36*5.5=1.98
S: __ 2.0 + 0.50(?!)*L __________________ 2 + 7.5*0.5 = 5.75 ____________ 0,60 __________ 0.54*7.5=4.05
O: __ 5.0 + 1.0 *L _____________________ 5 + 10*1.0 = 15 ______________ 1,00 __________ 0.90*10 = 9.0
Note that the European (NEM) weights are 'way lighter than the NMRA weights, and especially for TT (note to the calculations: 0,17 g/mm = 0.15 oz/in, etc); and NEM does not have a "starting" weight.
From one scale to the next, scale ratio > cubed = volume ratio: N>TT: 160/120 = 1.33 > 2.36; TT>HO: 120/87 = 1.38 > 2.63; HO>S: 87/64 = 1.36 > 2.51; S>O: 64/48 = 1.33 > 2.36
So, since the scale and volume ratio steps are almost equal from one scale to the next, and the weights should be proportional to volume, it looks (from the second column above) like the TT car is a bit heavy compared to N and HO, should be ~1.9 oz; and the S car is 'way light, should be ~7.8 oz (which it would be, if NMRA's per-inch multiplier were 0.75, which seems logical). And maybe the N-scale multiplier should be 0.25? That would bring the N car up to 1.25oz, making the NMRA TT weight look more nearly "in-line".
But comparing volumes to weights, from N to O = 36.8 by vol., 15.8 by wt., i.e., the smaller scales are proportionately twice as heavy? Example: scaling-by-volume a 40-foot prototype car, loaded wt = 140K lb => 0.55 oz for N, 20.3 oz for O! (and 1.30 oz for TT, 3.39 oz for HO, 8.54 oz for S) –
Makes you wonder how the early rule-makers came up with their formulas… possibly, using strict scaling, the larger-scale cars just seemed too heavy, and being worried about staying-on-track for the small scales, they decided to compress the range?
(End of my previous musings)
So: what "standard" to adopt? Personally, I feel heavier cars have better operating reliability and are less easily upset during uncoupling (assuming Kadee couplers), so my cars will follow the NMRA RP, at least as a goal.

Now - What are your preferences or (actual operating) experiences?