[max76]

The reference kilogram is estimated to have lost 50 micrograms over it's lifetime. I think it's safe to say that a reference gram would have lost a similar amount of mass. This would have caused a larger change in the definition of a gram/kilogram over time because the ratio of mass lost in the reference would have been greater.

Not much, but still.

[SilasX]

Why wouldn't the mass loss scale with the mass of the object? Why would it always be the same mass loss in absolute terms?

[max76]

It's hard to say because the mechanisms for the mass loss are not fully known. In fact, we are just guessing that mass loss actually occurred, and how much did occur. How would we have measured the mass change of the reference kilogram? By definition the reference kilogram was the most stable quantity of mass we could have generated at the time, and the most well protected quantity of mass for the duration of it's service as the reference kilogram.

I think mass loss is probably proportional with either the surface area or the surface area of the handled portion. This is just a hypothesis based on what I suspect the mechanism for mass loss is (friction from air resistance and being touched). Now that we have a more accurate way of measuring mass we could repeat the experiment of trying to preserve a quantity of mass, even doing different variations based on quantity, material, conditions, etc. Such an experiment would be relatively expensive and take a long time for good results, but maybe someone curious with funding will do it.

EDIT: The idea that we could have preserved a gram of material, while measuring it's mass every 40 years, for the last 130 years while it only lost 5 picograms (1/1000th of the estimated mass of the reference kilogram lost) is so crazy I can't believe it.

[SilasX]

Is the concept of a half-life similarly crazy to you?

[max76]

No, and radioactive decay doesn't account for the variances between the reference kilogram and it's sister kilograms we've observed.

[SilasX]

So it's an issue of "an event happens uniformly randomly across a mass" that seems unusual to you, and only applicable to radioactive decay?