I'm not so concerned with your safety - I'm concerned about less-experienced reloaders who might follow your examples. They are the ones who need to be watched over and guided, and if they mimick your example closely they may develop habits which can get them into trouble.
For the record, you have made quite a few basic mistakes in this single project which could spell trouble for less-experienced reloaders. For their benefit:
* First you assumed that the .25-20 has a greater case capacity than the .256 does.
* Second, you assumed that the Remington cases that Waters used have the same strength and hardness as the Winchester cases Sierra used.
* Third you assumed that the lots of powders used in the two tests made years apart were the same.
* Fourth you are surprised that the 256 starting load is "only" one grain over the max load Waters lists.
* Fifth, you assumed that the hard 87-grain Spitzer that Sierra used behaves the same as a soft 86-grain bullet (with a much longer shank) does in Waters tests.
* Sixth, you assumed that simply because you've used data from "hot" cartridges in older ones you won't get into trouble - you've "gotten away with it" before so it must be a safe practice.
* Seventh, you brag about how far you have exceed published loads, minimizing the potential for risk.
I'm not taking you to task because of what you are doing - you are the best judge of your own safety - but the example it sets for others with less experience is....questionable. Continue with your experiments - as I will after over thirty years of doing so - but please be a little more careful in how you present your ideas on unconventional techniques in an open forum read by new shooters and reloaders. Thanks for considering their safety in the future.
[ FWIW I've done a lot of experimenting with my BF handgun chambered in .25-20 over the past twelve years, and with its tight custom chamber (which supports the soft case well) it makes a fine high performance cartridge. But most of the shooting I do with it uses loads very close to published data - because it seems to shoot best there. ]