Click on a subject in the right-hand column or scroll down to read everything
Tolerance is defined as the total allowance for error in a specified figure, in other words, the sum of the absolute values of the positive and negative limits by which the nominal figure may be allowed to vary.
In practice, a low (negative) limit of -10% applied to a 10µF capacitor implies that it could have a capacitance as low as 9µF while a high (positive) limit of +10% implies it could rise as high as 11µF. The sum of the limits is 20% and the total range of the discrepancy from the nominal figure is 2µF - that is 20% of the nominal 10µF.
The entire scope of the limits -10% to +10% gives us a tolerance of 20%, but common usage has adopted the convention that the (two) limits of ±10% be classed (inaccurately) as a tolerance of 10%.
Limits of -0% +10% (or -10% +0%) equate to a tolerance of 10% (0+10), so here the tolerance is of the same magnitude as the limits.
As the word tolerance is that normally used for expressions like '±10%' we follow that custom, as we do where tolerance is implied, rather than specified with positive and negative limits (e.g. a stated tolerance of 10% implies ±10%). We therefore follow that convention in conversation and throughout this publication.