alt131 - 11:27 am on Sep 20, 2012 (gmt 0)

don't quote understand the reasoning for box-sizing being at risk when it has about 93% browser support (caniuse.com) and people (http://paulirish.com/2012/box-sizing-border-box-ftw/) are actively promoting it!

Thanks for drawing attention to that Paul. The padding-box value is at risk (which isn't even relevant here). Sorry team, tired brain late at night.

Lucy ... tables for text and quirks <sigh> :)

Are you certain about what you are seeing? On closer inspection I'm not so sure. Recalling outline is drawn around the edge of the border, so any variations are likely related to the interpretation of border-collapse (and for internal cells, maybe also stack order).

From a quick test on winxp, using ff14, op11, winSaf5 and ie8, using your code but with a border on the table, plus 10px solid border for the td, 5px dotted blue on hover.

With border-collapse:seperate all tested browsers have the outline outside the displayed cell border (and on top of the table border) - as expected.

With border-collapse:collapse, ff14 draws the outline outside the displayed cell border, all others draw the outline over the four sides of the border, starting at 5px across the border.

There is good argument that only 1/2 the border belongs to each cell, although the recs aren't explicit about the relationship between cell borders, but very clear about the relationship between cells and table borders:
17.6.2 The collapsing border model [w3.org]:

Borders are centered on the grid lines between the cells. ... The left border width of the table is half of the first cell's collapsed left border, and the right border width of the table is half of the last cell's collapsed right border. ...

I can't reproduce the top/left bottom/right variations, but I'd be interested if you are still seeing them given my thinking that stack order may also be at work here.