I have purchased several sets of dice from different online retailers. I did some salt-float tests on them recently and I'm concerned a good amount of my d20s are favoring a specific corner (all 3 like the 6/9 intersection for some reason).

I'm in a friendly campaign where we're not taking ourselves too seriously so "weighted dice" aren't necessarily a problem (especially with such low-rolling dice), but I still want to level the playing field of my dice.

I'm thinking of using a high-walled rolling tray and some textured padding, like a towel, so the bouncing from the walls and the friction from the towel would prevent the dice from "snapping" into desired positions.

How can I ensure my dice roll fairly despite their apparent bias, so as to not gain any unfair advantage or disadvantage, besides getting rid of them?

  • 6
    \$\begingroup\$ Have you actually tested to see if they are biased in their results? The float test just means they don't have uniform density, not necessarily that they roll unfairly. \$\endgroup\$ Sep 7, 2020 at 17:35
  • 6
    \$\begingroup\$ Related: How can I test whether a die is fair? \$\endgroup\$
    – Someone_Evil
    Sep 7, 2020 at 17:40
  • 1
    \$\begingroup\$ @ThomasMarkov I did spend some time rolling them on a flat surface and the float test seems to confirm they are rolling the numbers around the 6 and 9 area much more than normal unfortunately. \$\endgroup\$
    – diceofdoom
    Sep 7, 2020 at 17:51
  • 2
    \$\begingroup\$ I've revised this question to be more pointed with its request. “What would you do?” is an open-ended discussion question with no answers better or worse than any other and meets our Bad Subjective criteria. Instead, this would work best if you present a specific goal and ask how to accomplish it—then there are better or worse answers based on how well they assist you in accomplishing that goal. The goal here appears to be “make sure they roll decently fairly”, so I've updated the question to ask how to accomplish that. \$\endgroup\$ Sep 7, 2020 at 17:59
  • 1
    \$\begingroup\$ Does this answer your question? Dice Rolling Method Fairness \$\endgroup\$ Sep 7, 2020 at 18:12

2 Answers 2


You can't. Keep records of your roll results per die instead, and determine how severe the problem is

There isn't a whole lot you can do prospectively-- if you're pretty sure that the dice are not fair, you will have a hard time using them in a way that gets the same results as fair dice would. That is fundamentally what it means for a die to be biased. Some methods might compensate to varying degrees for specific manufacturing problems, but it strikes me as unlikely that you will find a method that exactly offsets your die's favor for 6s and 9s.

I can think of a couple of ad hoc things you could do to lessen systematic favoring of outcomes due to the bias, like having other players roll your dice in a rotation or shaking the dice in a cup before throwing rather than your hand. But these don't eliminate the problem; they just introduce more variation on other factors which could influence rolls (especially at the margins).

Something like a dice tower could help, but it won't necessarily cut out the bias and, crucially, it will introduce new systematic factors with an unknown influence on your roll results. It's possible, if not too likely, that the specific issue with your dice would be exacerbated by the tower, or might interact with it in some other systematic way that cuts against randomness. Inhibiting the rolling action, such as by rolling onto a towel, is likely to result in unworkable outcomes (like cocked dice) and may not counteract the die's issue very well, while also introducing new systematic factors that affect rolling.

What you can easily do is keep records of roll results to determine, after the fact, how severely your dice really are biased in practice. It's very easy to mis-estimate something like minor manufacturing issues with dice without empirical observation, and it can take a lot of rolls of a d20 to confidently assemble its distribution of roll outcomes. If the problem turns out to be pretty minor, it might not matter to you or your table. If it's severe enough to be an issue, at least you'll know and can stop using those dice.

Every other solution I can think of which would directly address your dice set's departure from fairness essentially cut the dice out altogether. The randomness a die introduces is based on its ability to physically move with a roughly equal chance of any face showing when it comes to rest. If that property is the one you don't trust, then you can't trust the dice.


Try a dice tower. The natural resting point for the dice is less relevant since the dice is stopped by the base of the tower rather than rolling to a stop.

If you feel like buying more dice though the try getting a set of these: https://darkelfdice.com/collections/gamescience-dice as they are designed to be properly random due to the way they are manufactured. (I have no affiliation with this product, or website other than owning a set of the dice).

  • 5
    \$\begingroup\$ I might also just mention the name of the brand without linking to a third-party reseller, since GameScience does have their own website where they sell the dice. (...Well, I'd actually recommend against them for reasons unrelated to the quality of their dice, but that's a separate point.) \$\endgroup\$
    – V2Blast
    Sep 7, 2020 at 21:22

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .