In that case, I think Firefox violates this. Your tabs are just gone when you could have sworn they were there a second ago...
In broad strokes, the time to accurately complete a task is proportional to the distance to an item and inversely proportional to the size of the item. So, having to scroll tabs increases the distance -- potentially a lot. But, constantly shrinking tabs reduces the target size. I think you'd have to do an amortized analysis on this one because upon a the creation of a new tab the Firefox model only makes that new tab harder to get to, while the Chrome model shrinks all tabs, making each harder to click.
But, I could buy that there's some threshold here where Chrome maintains a tab size that still doesn't appreciably increase error rate while improving distance by eliminating the tab scroll. I'd expect that minimal tab width to be much larger than Chrome's true minimum tab width. But, I'd be shocked if Google didn't run usability studies on this. I just haven't come across anything on my own.
General position is particularly helpful because, from left to right, it roughly represents "time opened", earliest to latest.
Firefox doesn't help me find the right tab just by including an 8 character width along with the favicon. Might as well just get rid of those 8 characters.
I'd love to read any study on the usability of the Chrome approach if anyone happens to know of one. Maybe something on SIGCHI?