17:39
@Mintx The rounding occurred in a term that defines the eccentricity of the ellipsoid. I wouldn't call it an "error" so much as a "difference." It affects most values after the eighth significant digit--less than 0.1 meters altogether.
There are other discrepancies between WGS84 and GRS80 that are not explainable by the choice of rounding. They are explained in section 22.1.3. A set of coordinates of "fundamental" points is needed in order to start surveying. Some differences in those coordinates arose because their calculations involved different sets of measurements, all obtained with some measurement error.
The discrepancies can reach up to one meter. It's not (easily) possible to say which datum is "correct" and which is not; they are equally correct on the face of it.
