In fairness they did not use a ruler or eyes based on the excerpts you quote they extracted exact coordinates of data from an svg format which if the svg was created correctly should at least give a non-biased dataset maybe with less precision than the source