Interesting when I get a result of 99.7% I am disappointed.
To each their own I guess.
And were we to use Petrie's measurement of 201.50 we would see the accuracy drop to 581.30 / 284.96 = 2.04 or accurate to 96.22
In all my work I have dealt with an accuracy of less than .5 of 1 percent and have basically discarded any results that fell outside these parameters I have always assumed that if we were being warned about something or shown something ... 96.22 just would not do. At least in my research that's the rule of thumb.
However that is not to forget that Scott gets his results, I believe, from scaling drawings. I do not believe he has used anything other than a measuring device on an image to get his results. This is inherently flawed as we all know who have tried to use just images to do our research.